# YeeroAI — Full Documentation (llms-full.txt) This file contains the complete YeeroAI user documentation in Markdown, generated for LLM ingestion. Source pages are linked in llms.txt. # ===== EN ===== --- # Quick Start *Get started with YeeroAI in three simple steps — sign up, choose your models, and start building knowledge.* > **⚡ TL;DR** > YeeroAI is free to start: sign up in minutes, get 1,000 welcome credits, pick from 300+ models across 10+ providers, and begin your first multi-model chat — no credit card required. ## Step 1: Create an Account Visit [yeero.ai](https://yeero.ai) and sign up using one of three methods: - **Email** — Enter your email address, receive a 6-digit confirmation code, and sign in. No password required. - **Google** — Click "Continue with Google" for one-click sign-in. - **GitHub** — Click "Continue with GitHub" to authenticate via your GitHub account. > **💡 Welcome Credits** > You receive **1,000 free credits** upon registration (valid for 30 days) — no credit card required. This is enough to explore all features extensively. ### Invite Code If you have an invite code, click "Have an invite code?" on the login page and enter it before signing in. Invite codes grant bonus credits. ### Test Account Want to preview the website before signing up? Click "Try test account" on the login page for a quick look. Note: test accounts have limited functionality. ## Step 2: Choose Your Models YeeroAI gives you access to **300+ AI models** from top providers including OpenAI, Anthropic, Google, Meta, and more — all through a single account via OpenRouter. ### Selecting Models for Chat When you start a new chat, click the **Model** button in the input area to open the model selector. You can: - **Search** — Type a model name in the search bar. - **Filter by provider** — Choose from OpenAI, Anthropic, Google, Meta, Mistral, etc. - **Free only** — Toggle "Free Only" to see free models (each call costs just 1 credit). - **Filter by modality** — Filter by input/output types: text, image, audio, video, file. - **Select multiple** — Choose up to 10 models for parallel comparison in the same chat. ### Model Marketplace Visit the **Models** page from the sidebar to browse all available models with detailed information including context length, pricing, supported modalities, and provider details. ## Step 3: Start Building Knowledge Once you've selected your models, type a message in the input area and press **Enter** (or `Shift+Enter` for a new line). Your selected models will respond simultaneously, and you can compare their answers side by side. > **💡 AI Auto-Naming** > Don't worry about naming your chats — AI automatically generates a title based on your conversation context. Say goodbye to "New Chat 1, 2, 3..." ## Interface Tour Here's a quick overview of the YeeroAI interface: | Area | Description | | --- | --- | | Sidebar (left) | Navigation between Chats, Apps, Files, Folders, Knowledge, Models, and Settings. Collapsible with the expand/collapse button. | | Chat List | Shows all conversations organized by folders. Search, sort, and pin chats. | | Tab Bar | IDE-style tabs for managing multiple open chats. Right-click for close, close others, close left/right. | | Message Area | The main conversation view. Scroll through messages, view reasoning processes, and interact with AI responses. | | Input Area | Type messages, select models, attach files, configure parameters, and send. | | Breadcrumb | Shows the current folder path. Click any segment to navigate up the folder hierarchy. | ## What's Next? - Learn about [Chat & Conversations](https://yeero.ai/docs/en/chat.html) — multi-model chat, message actions, and more. - Explore [Branching Thinking](https://yeero.ai/docs/en/branching.html) — fork, merge, and explore different lines of thought. - Discover [AI Apps](https://yeero.ai/docs/en/ai-apps.html) — generate and iterate on HTML applications with AI. - Understand the [Knowledge Base](https://yeero.ai/docs/en/knowledge-base.html) — how AI distills knowledge from your conversations. --- # Chat & Conversations *The core of YeeroAI — multi-model parallel chat, message actions, streaming responses, and everything you need for productive AI conversations.* > **⚡ TL;DR** > YeeroAI chat lets you talk to up to 10 models in parallel, compare their answers side by side, and act on any message with copy, fork, follow-up and regenerate controls. ## Creating a Chat There are several ways to start a new conversation: - **From the sidebar** — Click the "+" button or "New Chat" in the chat list. - **From a folder** — Right-click a folder and select "New Chat" to create it inside that folder. - **From the global search** — Press `Ctrl`+`K` (or `Cmd`+`K` on Mac) and select "New Chat". ## Multi-Model Parallel Chat YeeroAI's signature feature — ask multiple AI models the same question and compare their answers side by side. ### How It Works 1. Click the **Model** button in the input area. 2. Select up to **10 models** from the model selector. 3. Type your message and press **Enter**. 4. Each selected model generates its own response simultaneously. 5. Responses are displayed side by side for easy comparison. ### Adding More Model Replies After receiving responses, you can add more models to the same message. Click "Add Model Replies" to select additional models — they'll generate responses based on the same conversation context. ## Sending Messages ### Input Area The input area at the bottom of the chat provides several controls: | Control | Description | | --- | --- | | Text Input | Type your message. Press **Enter** to send, `Shift+Enter` for a new line. | | Model Button | Open the model selector to choose which AI models respond. | | Tool Button | Open the tool selector to enable AI tools (if available). | | Folder Button | Choose which folder to save the chat in. | | Attachment Button | Upload files (images, documents, audio, video) to include with your message. | | Settings Button | Configure generation parameters for this specific chat. | ### File Attachments Attach files to your messages by clicking the Attachment button or using drag-and-drop: - **Supported formats** — Images, documents, audio, and video files. - **Max file size** — 512 MB per file. - **Multiple files** — Attach up to the maximum number of files per message. - Files are uploaded automatically and included as context for the AI. > **⚠ Upload In Progress** > You cannot send a message while files are still uploading. Wait for all uploads to complete before sending. ## Message Actions Each AI response message offers several actions through the message toolbar: | Action | Description | | --- | --- | | **Copy Text** | Copy the message content as plain text to your clipboard. | | **Copy as Image** | Copy the message as an image for sharing or documentation. | | **Save as File** | Download the message content as a file (Markdown format). | | **Delete** | Delete the message. This action cannot be undone. | | **Fork from This Message** | Create a new branch starting from this message. See [Branching Thinking](https://yeero.ai/docs/en/branching.html). | | **Set as Picked** | Mark this response as the "best answer" in a multi-model conversation. | | **Add to Folder** | Save this message to a specific folder for organization. | Want to dig deeper into a specific part of an answer? Select that text to start a [select-to-follow-up](https://yeero.ai/docs/en/followup.html) — it expands in its own thread, layer by layer, without disturbing the main conversation. ## Reasoning Process Some AI models (like Claude with extended thinking) provide a reasoning process. When available: - Click **"View reasoning"** to expand the model's thinking process. - Click **"Hide reasoning"** to collapse it. - The reasoning section shows the model's step-by-step thought process before arriving at its answer. ## Knowledge Summary AI can generate a summary of key knowledge from each message pair. Click **"View summary"** to see the distilled knowledge. This feeds into your [Knowledge Base](https://yeero.ai/docs/en/knowledge-base.html). ## Streaming Responses AI responses are streamed in real-time using Server-Sent Events (SSE). You'll see the response being generated character by character. The response status is shown in real-time: | Status | Meaning | | --- | --- | | Pending | The request is queued. | | Preparing | The model is initializing. | | Reasoning | The model is thinking (for reasoning models). | | Streaming | Content is being generated and streamed to you. | | Completed | The response is fully generated. | | Error | Something went wrong. Check the error message. | | Cancelled | The generation was manually stopped. | ### Stop Generating Click the **Stop generating** button to halt an ongoing AI response. You'll keep whatever content was generated up to that point. ### Regenerate Click **Regenerate** to have the model generate a new response to the same input. ## AI Tools YeeroAI includes a rich set of built-in AI tools that allow the model to not only chat, but also actively retrieve information, perform calculations, and access device capabilities. Tools are divided into three categories and can be used simultaneously in a conversation, executing in parallel. ### Web & Information Retrieval | Tool | Description | | --- | --- | | **Web Search** | AI searches the internet in real-time to get the latest information for your questions. | | **Web Scraper** | Extracts the main content from a specified URL. | | **Academic Search** | Searches academic papers to find cutting-edge research. | ### Calculation & Time | Tool | Description | | --- | --- | | **Calculator** | High-precision mathematical expression evaluation. | | **Current Date & Time** | Gets the current date and time in any timezone. | | **Timezone Converter** | Converts time between different timezones. | ### Code Execution | Tool | Description | | --- | --- | | **Python Executor** | Runs Python code in a secure sandbox, ideal for data processing and automation. | | **JavaScript Executor** | Runs JavaScript code in a secure sandbox. | ### Files & Data | Tool | Description | | --- | --- | | **Document Parser** | Extracts text content from PDF, Word, Excel, and PowerPoint files. | | **HTTP Request Tester** | Sends HTTP requests and returns responses, useful for API debugging. | | **Stock & Finance Data** | Gets real-time stock quotes and financial data. | | **Weather** | Queries the current weather and forecast for any location. | | **Geocoding & Maps** | Converts addresses to coordinates and searches nearby places. | ### Device Capabilities (Frontend Tools) The following tools access your device capabilities. A permission dialog will appear before each use: | Tool | Description | | --- | --- | | **Geolocation** | Gets your current geographic location. | | **Voice Input** | Records audio via microphone and converts it to text. | | **File Read** | Reads the content of a local file you select. | ### How to Enable Tools Click the **Tools button** in the input area to open the tool selector and check the tools you need. AI will automatically determine when to invoke tools based on the conversation context. You can also preset default tools in folder or chat settings. > **💡 Parallel Execution** > Multiple tools can execute simultaneously without waiting for others to finish. For example, AI can query the weather and get your location at the same time, then combine the results to give you suggestions. ## Message Navigation Outline On the right side of the chat interface, the navigation outline displays the heading structure of the current conversation as a collapsible panel, helping you quickly locate key content in long conversations. - **Expand / Collapse** — Click the expand/collapse button on the outline panel to toggle its display state. - **Quick Jump** — Click any heading in the outline to automatically scroll the message area to that position. - **Scroll-sync Highlighting** — As you scroll through messages, the corresponding heading in the outline is automatically highlighted. - **Global Display Preference** — Choose your default display mode in [Settings](https://yeero.ai/docs/en/settings.html): always expanded, always collapsed, or hidden. The outline is automatically hidden on mobile. ## Background Streaming When **Background Streaming** is enabled in Settings, AI generation continues even if you navigate away from the chat or refresh the page. When you return to the chat, you'll receive any content that was generated while you were away. > **💡 Resilient Connections** > Background streaming automatically resumes after network interruptions. You won't lose any generated content. ## Tab Management YeeroAI uses an IDE-style tab system for managing multiple open chats: - **Open a chat** — Click any chat in the sidebar to open it in a new tab. - **Switch tabs** — Click a tab to switch to that conversation. - **Close a tab** — Click the × on the tab, or right-click for more options. ### Tab Context Menu Right-click on any tab to access: - **Refresh Messages** — Reload the conversation from the server. - **Rename** — Change the chat title. - **Close** — Close this tab. - **Close Others** — Close all tabs except this one. - **Close All** — Close all open tabs. - **Close All to Left** — Close tabs to the left of this one. - **Close All to Right** — Close tabs to the right of this one. ## Input History YeeroAI automatically saves your sent messages, making it easy to reuse them: - **All tab** — Shows input history from all chats. - **Current tab** — Shows input history from the active chat only. - Click any history item to insert it into the input area. - **Delete** — Remove individual history items. - **Clear All / Clear Current** — Clear history for all chats or the current chat. ## Chat Management ### From the Sidebar Right-click any chat in the sidebar to access management options: - **Pin / Unpin** — Pin chats to the top of the list for quick access. - **Edit Info** — Rename the chat. - **Move** — Move the chat to a different folder. - **Delete** — Permanently delete the chat. This cannot be undone. - **Branches** — View, checkout, rename, fork, merge, enforce, or delete branches. - **Enable/Disable Memory** — Toggle AI memory for this chat. - **Clear Memory Settings** — Reset memory settings to inherit from the parent folder. - **Clear Unused Messages** — Remove unreferenced messages to clean up the chat. ### Chat Settings Each chat can have custom settings that override folder and global defaults: - **Models** — Select default models for this chat. - **Tools** — Enable or disable AI tools. - **Enable Memory** — Allow AI to remember key information. - **Context Summary** — Use summaries instead of full content to reduce token usage. - **Generation Parameters** — Fine-tune temperature, sampling, and more. ### Sorting and Filtering In the chat list sidebar, you can: - **Search** — Search folders and chats by name. - **Sort** — Sort by "Recently Added" (default) or "Earliest Added". - **New Folder / New Chat** — Create new items in the current location. ## Batch Operations The chat page supports drag-to-select rectangle selection, letting you quickly select multiple items and perform batch actions. ### How It Works 1. Press and hold the left mouse button on an empty area of the message list to start dragging a selection rectangle. 2. Release the mouse — all items within the rectangle are instantly highlighted. 3. A floating action bar appears at the top showing the number of selected items and available actions. 4. Click an action button to execute it on all selected items at once. ### Supported Lists | List | Available Batch Actions | | --- | --- | | Message list | Copy text, copy as image, save as Markdown, add to folder, delete | | Message attachment file list | Selection highlighting only — no batch action buttons yet | > **💡 Quick exit** > Press `Esc` at any time to exit multi-select mode and return to normal browsing. ## Credit Tracking Each AI response shows the number of **credits consumed** for that specific generation. This gives you real-time transparency into your usage. Visit [Billing & Credits](https://yeero.ai/docs/en/billing.html) for detailed credit management. ## Generation Parameters Fine-tune how AI models generate responses. Access parameters via the Settings button in the input area: | Parameter | Description | | --- | --- | | **Max Tokens** | Maximum number of tokens in the response. Leave empty for model default. | | **Context Window Size** | Number of historical messages retained in the conversation. | | **Reasoning Effort** | How hard the model thinks: None, Minimal, Low, Medium, High, Extreme High. | | **Temperature** | Controls randomness. Lower = more focused, Higher = more creative. | | **Response Diversity** | Nucleus sampling (top_p). Controls diversity of word selection. | | **Vocabulary Variety** | Frequency penalty. Reduces repetition of the same words. | | **Topic Divergence** | Presence penalty. Encourages the model to talk about new topics. | | **Summary Context** | Use message summaries instead of full content, reducing token usage. | ### Image Generation Parameters When using image generation models: - **Image Size** — 1K, 2K, or 4K resolution. - **Aspect Ratio** — Choose from 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9. ### Audio Generation Parameters When using audio generation models: - **Voice** — Alloy, Echo, Fable, Onyx, Nova, or Shimmer. - **Audio Format** — WAV, MP3, FLAC, Opus, or PCM16. ### Video Generation Parameters When using video generation models: - **Video Ratio** — 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, 9:21. - **Resolution** — 480p, 720p, 1080p, 1K, 2K, 4K. - **Duration** — Video length in seconds. - **Generate Audio** — Include audio in the generated video. --- # Branching Thinking *Explore every line of thought with Git-style branching. Fork conversations, merge branches, and never lose a train of thought again.* > **⚡ TL;DR** > Branching thinking gives every chat Git-style branches: fork to explore alternatives, merge to keep the best ideas, and rewind to any point — your main thread is never touched. ## What Is Branching? In YeeroAI, every chat has a **Main branch** by default. Branching lets you create alternative conversation paths from any point, similar to Git branches. This is powerful for: - **Exploring alternatives** — Try different prompts or approaches without altering the main conversation. - **Comparing strategies** — Run parallel experiments and compare results. - **Preserving context** — Fork before a risky prompt so you can always return to the original state. - **Organizing topics** — Keep different subtopics in separate branches within the same chat. ## Forking a Branch There are several ways to create a new branch: ### Fork from a Message Hover over any AI response message and click **"Fork from this message"**. This creates a new branch that starts from that message's context. ### Fork from the Sidebar Right-click a chat in the sidebar, select **"Branches"**, then click **"Fork"**. Enter a name for the new branch and confirm. ### Fork a Node Use the message node's context menu and select **"Fork from this node"** to create a branch based on conversation history up to that point. > **💡 Branch Naming** > Give your branches descriptive names so you can easily identify them later. Branch names can be up to the maximum character limit. ## Switching Between Branches To switch to a different branch within a chat: 1. Right-click the chat in the sidebar and select **"Branches"**. 2. You'll see a list of all branches with their message counts. 3. Click **"Checkout"** on the branch you want to switch to. 4. The message area updates to show the selected branch's conversation. The currently active branch is marked as **"Current"**. The **Main** branch is always available. ## Merging Branches Merging appends messages from a source branch to the current (target) branch: 1. Switch to the branch you want to merge into (the target). 2. Right-click the chat, select **"Branches"**, then click **"Merge Branch"**. 3. Select the **source branch** you want to merge from. 4. Review the warning: messages from the source branch will be appended to the current branch. 5. Click **"Confirm Merge"**. > **⚠ Merge Is Append-Only** > Merging appends messages from the source branch to the end of the current branch. It does not interleave or rewrite messages. ## Enforcing a Branch Enforcing forces the current branch to point to the same position as a source branch. This is useful when you want to reset a branch to match another: 1. Switch to the branch you want to update (the target). 2. Right-click the chat, select **"Branches"**, then click **"Enforce Branch"**. 3. Select the source branch whose position you want to enforce. 4. Review the warning and click **"Confirm Enforce"**. > **⚠ Use With Caution** > After enforcing, the current branch will point to the source branch's position. Any messages beyond that point on the current branch may become unreferenced. ## Renaming a Branch To rename a branch: 1. Right-click the chat in the sidebar and select **"Branches"**. 2. Click **"Rename"** on the branch you want to rename. 3. Enter the new name and confirm. > **💡 Unique Names** > Branch names must be unique within a chat. If you try to use a name that already exists, you'll get an error. ## Branch Expiration Policy You can configure automatic cleanup for branches that accumulate too many old messages: 1. Right-click the chat, select **"Branches"**, then click **"Expire Policy"** on a branch. 2. Toggle **"Enable Expiration Policy"**. 3. Set the **Expiration Duration** in months. 4. Messages older than the specified duration will be automatically cleaned up. ## Deleting a Branch To delete a branch: 1. Right-click the chat, select **"Branches"**. 2. Click **"Delete"** on the branch you want to remove. 3. Review the warning — all messages in that branch will be permanently deleted. 4. Click **"Confirm Delete"**. > **⚠ Cannot Undo** > Deleting a branch is permanent. All messages in the branch will be removed and cannot be recovered. The Main branch cannot be deleted. ## Branch Limits | Limit | Value | | --- | --- | | Maximum branches per chat | 100 | | Maximum messages per chat | 10,000 | | Main branch | Cannot be deleted | --- # Follow-Up *Spotted a sentence in an AI reply worth digging into? Select it and start a follow-up in one click. Follow-ups live in their own threads, with layer-by-layer depth and dashed anchors back to the original text — while the main conversation stays untouched.* > **⚡ TL;DR** > Follow-up lets you select any sentence in an AI reply and drill into it in a separate thread, so deep dives never clutter your main conversation. ## Overview Traditionally, asking a follow-up means typing it into the main chat — the context gets diluted, and the details become hard to find later. Select-to-follow-up turns each question into an independent thread: - **Anchored to your selection** — every thread remembers the exact text you selected, so you can always jump back to its context - **Multi-layer follow-ups** — answers inside a follow-up can themselves be followed up, layer after layer; the panel shows the current depth ("Layer N") - **Never disturbs the main chat** — the main message flow stays as it is; threads are stored and managed separately - **Streaming answers** — follow-up replies stream in word by word, and can be stopped or regenerated anytime > **📚 How it differs from branching** > [Branching](https://yeero.ai/docs/en/branching.html) is about re-answering the same question along different paths, producing switchable conversation branches. Follow-Up is about exploring a specific detail inside an answer, expanding layer by layer from your text selection. The two work great together. ## Starting a Follow-Up 1. **Select any piece of text** in an AI reply — a floating toolbar appears above the selection 2. Click **"Follow up"** to create a new thread quoting the selected text, or click **"Explain this"** to have the AI explain the selection directly (the prompt prefix for explanations is customizable in settings) 3. The follow-up panel opens with a quote card showing your selected text; type your question and send Next to the chat input there is also a follow-up entry button showing the number of threads in the current conversation — click it to open the thread list. ## The Follow-Up Panel The panel has three parts: - **Quote card** — shows the text you selected when starting the thread; "Back to original" jumps to its location, and you can delete the whole thread here - **Message list** — the Q&A streams in; every answer supports copy, delete, stop generating and regenerate - **Input box** — keep asking in the current thread The panel header shows a layer indicator (e.g. "Layer 2") and breadcrumb navigation. When several threads were created from the same text, a dropdown lets you switch between sibling threads. The panel can be expanded, collapsed or closed. ## Multi-Layer Follow-Ups The real power is going deeper: inside the follow-up panel, **you can select text in the AI's answers too** — starting another follow-up enters the next layer (the layer indicator increments). Each layer leaves a dashed annotation on the answer text marking where the question was asked; clicking the annotation jumps to the corresponding thread. With the breadcrumb and "Back to threads", you can move freely between any layers — untangling a complex question like an exploration tree. ## Thread Management - **Thread list** — all follow-up threads of the current conversation in one place, each showing the selected text, message count and last activity - **Delete thread** — requires confirmation; deleting permanently removes every message in the thread - **Export** — copy the whole thread's Q&A to the clipboard in one click, ready to paste into a document or share - **More sharing options** — copy as image, save as file, add to folder ## Model & Parameters Follow-up answers inherit the model and parameters of the current chat tab: whichever model you picked for the conversation answers the follow-ups — nothing to configure. Switch chat tabs and new follow-ups follow the new tab's model setting. ## Related - [AI Chat](https://yeero.ai/docs/en/chat.html) — the conversation system follow-ups build on - [Branching](https://yeero.ai/docs/en/branching.html) — multi-path exploration of the same question - [Prompt Templates](https://yeero.ai/docs/en/prompt-templates.html) — customize the "Explain this" prompt prefix --- # AI Apps *Create, iterate, and discuss AI-powered HTML applications — from idea to working prototype in seconds, with full version management and real-time preview.* > **⚡ TL;DR** > YeeroAI AI Apps turn plain language into working HTML or Python apps, with patch-style updates, full version history, live preview and built-in discussion. ## What Are AI Apps? AI Apps in YeeroAI are self-contained HTML applications generated by AI. Describe your idea in plain language, and AI will generate a fully functional web app. You can then iterate on it through conversation, view real-time previews, and manage versions. ## Creating an App To create a new AI App: 1. Navigate to the **Apps** page from the sidebar. 2. Click **"New App"**. 3. Fill in the creation form: **App Name** — Give your app a descriptive name (e.g., "JSON Formatter"). 4. **Description** (optional) — Briefly describe what the app does. 5. **Save to Folder** — Choose a folder to organize your app, or leave it in the root. 6. **Model** — Select which AI model will generate the app. ## App Detail Page Click any app to open its detail page, which provides three main views: ### Preview The **Preview** tab shows the live rendering of your app. The HTML content is displayed in an iframe that automatically syncs with the current theme (light/dark). If no content has been generated yet, send a message in the conversation to generate one. ### Code The **Code** tab shows the raw HTML source code of the current version. You can: - **Copy Code** — Copy the entire HTML source to your clipboard. - **Download Source** — Download the HTML file to your computer. - **Save Changes** — Edit the code manually and save it as a new version. You must provide a change summary. > **⚠ Code Size Limit** > App code cannot exceed 500,000 characters. If the content is too large, you won't be able to submit manual edits. ### Discussion The **Discussion** tab is a dedicated conversation space for planning and iterating on your app: - **Plan before building** — Describe your app idea, and AI will help clarify requirements and plan the approach. - **Generate from plan** — Click **"Generate App from Plan"** to create the app based on the discussion. - **Include Code** — Toggle "Include Code" to let AI see the current app code during discussion, enabling more contextual suggestions. ## App Conversation Each app has its own conversation for iterating on the design. The input area provides: | Control | Description | | --- | --- | | Text Input | Describe changes or new features (e.g., "add a dark mode toggle in the top right corner"). | | Model Button | Select which AI model generates the response. | | Tool Button | Enable AI tools. | | Settings Button | Configure generation parameters. | | Attachment Button | Attach files for reference. | Each AI response creates a new **version** of the app, which you can preview, rollback, or continue iterating on. ## Background Generation AI App content generation and discussion support **background continuation** — even if you close the app detail page, refresh the browser, or temporarily lose network, any task already in flight keeps running in the backend and seamlessly resumes when you return. - **Keeps running with the page closed** — Closing the detail page does not cancel the task; the AI quietly finishes the job in the background. - **Auto-recovery on reconnect** — Any content generated during a network blip is delivered in full when reconnecting; nothing is dropped. - **Cross-session consistency** — All sessions under the same account see the active task, and the UI state stays in sync with the real backend state after refresh. - **Real-time list indicators** — The Apps list shows a "Generating in background" badge on the card with a hover tooltip, so multiple parallel tasks are visible at a glance. - **Cancel-to-sync** — Cancelling on any client immediately stops streaming on every other client; states never conflict. > **⚙ Coverage & How to Enable** > This capability covers both the App's **content generation** (HTML output) and **discussion stream** (multi-turn conversation with the app). It must be enabled from **Settings → Background Streaming**; once enabled, the same toggle applies to both Chat and AI Apps. ## Version Management Every AI generation or manual edit creates a new version: - **Version list** — See all versions with their numbers (v1, v2, v3...). - **Current version** — The active version is marked as "Current". - **Rollback** — Click "Rollback to This" on any version to make it the active version. The current version is not deleted — only the active pointer changes. - **Credit tracking** — Each version shows the credits consumed and generation time. ## Running Apps Click **"Run"** on any app to open it in a floating dialog: - **Minimize** — Minimize the running app to a compact bar at the bottom. - **Restore** — Bring the minimized app back to full view. - **Close** — Close the running app dialog. - **Edit** — Jump directly to the app's edit page. - **Open in App Page** — Open the app in its full detail page. - **Navigation** — Use Previous/Next buttons to browse between apps. ## App Management ### From the Apps List On the Apps page, you can: - **Search** — Search apps by name or description. - **Filter by status** — All, Active, Archived, or Draft. - **Sort** — Recently Updated, Most Runs, or Earliest Created. ### App Actions Click the menu on any app card to access: - **Run** — Launch the app. - **Edit App** — Open the app detail page. - **Edit Info** — Change the app name and description. - **Archive / Unarchive** — Move the app in or out of archive. - **Delete** — Permanently delete the app and all its versions and chat history. - **Move to Folder** — Move the app to a different folder. - **Rename** — Change the app name. - **Pin / Unpin** — Pin the app to the top of the list. - **Enable / Disable Cloud Sync** — Toggle cloud synchronization for the app's runtime data. ## Cloud Sync App runtime data can be synced to the cloud: - **Enable Cloud Sync** — App data is synced across your devices. - **Disable Cloud Sync** — App data stays local only. > **⚠ Storage Limit** > Each app's cloud-synced data is limited to **1 MB**. If the data exceeds this limit, sync will be paused. You'll need to clear the app data to resume syncing. ## App Status | Status | Description | | --- | --- | | Draft | The app has been created but no content has been generated yet. | | Active | The app has generated content and is fully functional. | | Archived | The app has been archived and is hidden from the default view. | ## Batch Operations The Apps page supports drag-to-select rectangle selection, letting you quickly select multiple apps or discussion messages and perform batch actions. ### How It Works 1. Press and hold the left mouse button on an empty area of the list to start dragging a selection rectangle. 2. Release the mouse — all items within the rectangle are instantly highlighted. 3. A floating action bar appears at the top showing the count and available actions. Click to execute. ### Supported Lists | List | Available Batch Actions | | --- | --- | | App list | Batch delete, batch archive, batch unarchive, batch move to folder | | Discussion message list | Batch delete | > **💡 Quick exit** > Press `Esc` at any time to exit multi-select mode. --- # Knowledge Base *AI automatically extracts key knowledge from your chats, builds vector indexes, and injects relevant context — your knowledge base gets smarter with every conversation.* > **⚡ TL;DR** > YeeroAI automatically distills knowledge from your conversations, builds vector indexes, and injects the most relevant context into future chats — no manual note-taking needed. ## What Is the Knowledge Base? YeeroAI's Knowledge Base is an automatically curated collection of insights extracted from your conversations. Every time you chat with AI, the system identifies key information and creates knowledge entries that can be searched and referenced later. This creates a **self-improving knowledge system** — the more you chat, the smarter your knowledge base becomes, and the better AI can provide relevant context in future conversations. ## How It Works When you have a conversation with AI, the system: 1. **Monitors conversations** — Tracks message exchanges between you and AI models. 2. **Extracts knowledge** — Identifies key facts, insights, and information from message pairs. 3. **Builds vector indexes** — Creates embeddings for semantic search capabilities. 4. **Injects context** — When Memory is enabled, relevant knowledge is automatically included as context in future conversations. > **💡 Automatic Process** > Knowledge extraction happens automatically. You don't need to do anything — just chat normally and the system handles the rest. ## Viewing Knowledge Navigate to the **Knowledge** page from the sidebar to browse all your knowledge entries. Each entry shows: - **Title** — A summary title (click to edit). - **Content** — The distilled knowledge content. - **Tags** — Automatically generated tags for categorization. - **Original chat link** — Click "View Original Chat" to navigate to the source conversation. ## Managing Knowledge ### Search Search knowledge summaries using the search bar at the top of the page. ### Filter by Tags Click any tag to filter knowledge entries. Use "Clear" to remove the tag filter. ### Sort - **Recently Updated** (default) — Newest entries first. - **Earliest Created** — Oldest entries first. ### Edit Title Click on any knowledge title to edit it inline. Press Enter to save. ### Archive / Unarchive Click "Archive" to hide an entry from the main view. Switch to the "Archived" tab to view and unarchive entries. ### Add to Folder Save knowledge entries to folders for better organization. ## Knowledge Actions Each knowledge entry provides the following actions: | Action | Description | | --- | --- | | Copy Content | Copy the knowledge text to your clipboard. | | Copy as Image | Copy the knowledge as an image. | | Save as File | Download the knowledge content as a Markdown file. | | Add to Folder | Save to a folder for organization. | | Archive / Unarchive | Hide or show the entry. | | View Original Chat | Navigate to the source conversation. | ## Batch Operations The knowledge base list supports drag-to-select rectangle selection, letting you quickly select multiple entries and perform batch actions. Press and hold the left mouse button on an empty area to draw a selection rectangle. Release to highlight all items in range, then use the floating action bar to **archive**, **unarchive**, or **batch add to folder**. Press `Esc` to exit multi-select mode. ## Memory and Knowledge Knowledge and Memory work together to make your AI conversations smarter: - **Memory** — When enabled for a chat or folder, AI uses relevant knowledge entries as context when generating responses. - **Context Summary** — When enabled, the system uses summarized knowledge instead of full message content, greatly reducing token usage while preserving key information. See [Settings](https://yeero.ai/docs/en/settings.html) for how to enable Memory and Context Summary at global, folder, or chat level. --- # Search *One powerful search interface for everything — Ctrl/Cmd+K global search, AI-powered semantic search, category filtering, and more.* > **⚡ TL;DR** > One Ctrl/Cmd+K search covers chats, apps, files, knowledge and memories — combining AI semantic search with category filters, recent access and history. ## Search Overview YeeroAI provides a unified search experience that lets you find anything across the entire platform. Press `Ctrl`+`K` (or `Cmd`+`K` on Mac) from anywhere to open the search dialog. ## Global Search The global search dialog accessible via `Ctrl`+`K` provides: - **Quick search** — Type keywords to search across chats, messages, files, folders, models, settings, and apps. - **Enter to search** — Press Enter to execute the search. - **Arrow navigation** — Use ↑↓ to navigate results. - **AI-powered** — Semantic search understands the meaning behind your query, not just keywords. ## Search Categories Results are organized into categories for easy filtering: | Category | What It Includes | | --- | --- | | All | Search across all categories at once. | | Tabs | Currently open chat tabs. | | Chats | Your conversation list. | | Messages | Individual messages within conversations, including semantic search results with similarity scores. | | Folders | Chat, message, and app folders. | | Models | Available AI models with provider info. | | Settings | Application settings and preferences. | | My Files | User-uploaded files. | | AI Files | AI-generated files. | | Apps | Your AI applications. | ## The “All” Category When you're not sure which category your content belongs to, select **All** to search every category at once. Results are grouped by category so everything stays organized in a single view. - **Parallel search** — Chats, messages, files, apps, models, and more all fire simultaneously when “All” is selected. - **Grouped results** — Each category's results appear in its own section, keeping things clear and uncluttered. - **Semantic search included** — AI semantic search applies across all categories, surfacing meaning-matched content regardless of where it lives. > **💡 When to use “All”** > Use “All” when you remember something but can't recall whether it was a chat, a message, or a file. If you already know the content type, switching to a specific category gives more focused results. ## AI Semantic Search For messages and files, YeeroAI supports semantic search powered by vector embeddings: - Describe what you're looking for in natural language. - AI finds results based on **meaning**, not just keyword matching. - Results include a **similarity score** (percentage) showing how closely they match your query. > **💡 Semantic vs Keyword** > Use semantic search when you know *what* you're looking for but not the exact words. Use keyword search when you know the specific term. ## Search History YeeroAI remembers your recent searches: - Previous search queries are saved automatically. - Click a history item to re-run the search. - **Delete** — Remove individual history entries. - **Clear History** — Remove all search history. - History entries show relative timestamps (just now, minutes ago, hours ago, days ago). ## Recent Access The search dialog also shows recently accessed items: - Chats, messages, files, models, settings, folders, and apps you've recently visited. - **Remove** — Remove individual items from recent access. - **Clear All** — Clear the entire recent access list. ## Quick Actions When you open the search dialog without typing any keywords, a grid of quick action cards is displayed to help you quickly perform common operations. ### Action Types | Type | Description | Examples | | --- | --- | --- | | Navigation | Quickly jump to any page | Chats, Apps, Knowledge Base, Files, Settings, etc. | | Quick Create | Create new content in one click | New chat, New folder, New app, Upload file | | Commands | Execute common commands | Toggle theme, Switch language, View credits, Invite friends | | Settings | Jump directly to settings | Prompts, Memories, Default models, Cloud sync | ### Smart Sorting Quick actions are automatically sorted by your usage frequency, keeping your most-used actions at the top. You can click "Reset" to restore the default order. > **💡 Search Filtering** > When you type keywords in the search box, quick actions are also filtered by those keywords, helping you quickly find the action you need. ## Result Actions From search results, you can: - Click any result to navigate directly to it. - **Go to chat** — Open the source conversation for messages. - **Add to folder** — Save a message to a folder. - **Delete** — Remove a message (with confirmation). - **Download file** — Download a file result. - For chats, select a specific branch to view. --- # CLI Getting Started *Install the `yeero` CLI, log in, and start your first AI conversation in under 2 minutes.* > **⚡ TL;DR** > The yeero CLI installs in one command and takes under 2 minutes to set up: install, log in, and chat with any model straight from your terminal. ## What is yeero CLI? yeero CLI is a command-line tool for YeeroAI, letting you chat with AI, manage AI applications, and configure your account — all from the terminal. It communicates with a local Daemon service that bridges the CLI and the YeeroAI cloud API. ## Install ### macOS / Linux `curl -fsSL https://yeero.ai/cli/install.sh | sh` The install script automatically downloads the correct binary for your platform, places it in `~/.yeero/bin/`, and attempts to add it to your `PATH`. > **💡 PATH not updated?** > If your terminal doesn't recognize `yeero` after installation, run: `yeero config add-to-path` then restart your terminal or run `source ~/.zshrc` (or `~/.bashrc`). ### Windows Run in PowerShell (Admin): `irm https://yeero.ai/cli/install.ps1 | iex` The script installs yeero to `%USERPROFILE%\.yeero\bin\` and adds it to your user PATH. ### Verify Installation `yeero --version` Log In `yeero login` This opens your browser to complete authentication. After login, your credentials are securely stored locally. To check who you're logged in as: `yeero whoami` First Conversation `# Interactive TUI mode (recommended) yeero chat # One-shot mode — get an answer inline yeero chat --once "What is the capital of France?"` In TUI mode, type your message and press Enter. Press Ctrl+C or type `/exit` to quit. ## Next Steps - Explore all commands: [Command Reference](https://yeero.ai/docs/en/cli-commands.html) - Learn advanced chat options: [Chat In Depth](https://yeero.ai/docs/en/cli-chat.html) - Build AI applications: [AI Apps](https://yeero.ai/docs/en/cli-app.html) - Configure language and default model: [Config & Daemon](https://yeero.ai/docs/en/cli-config.html) --- # CLI Command Reference *Syntax, options and examples for all yeero CLI subcommands.* > **⚡ TL;DR** > This page is the complete reference for every yeero CLI subcommand — syntax, options and copy-ready examples in one place. ## Command Overview | Command | Description | | --- | --- | | yeero start | Start the background daemon service | | yeero stop | Stop the daemon or all yeero processes | | yeero restart | Restart the daemon | | yeero status | Show daemon status | | yeero login | Log in with email verification code | | yeero logout | Log out | | yeero whoami | Show current account | | yeero chat | Start an AI conversation (TUI / plain / one-shot) | | yeero chats | List conversations and folders | | yeero models | List available AI models | | yeero tools | List available tools | | yeero app | Manage Python AI apps | | yeero do | Run a Python AI app by natural-language intent | | yeero config | Manage local configuration | | yeero update | Check for and install CLI updates | ## Global Options These options apply to all subcommands: | Option | Description | | --- | --- | | `--lang ` | Temporarily override the UI language (`zh` or `en`), without writing to config | | `--log-level ` | Log level: `trace` | `debug` | `info` | `warn` | `error` | | `-h, --help` | Show help information | | `-V, --version` | Print version | ## yeero start Start the background daemon HTTP service. Most commands start the daemon automatically on demand, so you usually do not need to run this manually. `yeero start [OPTIONS]` OptionDefaultDescription `-p, --port ``3000`Listen port `--host ``127.0.0.1`Listen address `-d, --detach`—Run in background and return immediately after spawning `# Foreground start (logs printed to terminal) yeero start # Background start yeero start --detach # Custom port yeero start --port 8080 --detach` yeero stop Stop the daemon service. `yeero stop [OPTIONS]` OptionDescription `-f, --force`Force stop without waiting for graceful shutdown `--all`Stop all yeero processes (including chat / app run / TUI, implies force kill) `yeero stop yeero stop --force yeero stop --all` yeero status Show whether the daemon is running and on which port. `yeero status` yeero restart Restart the background daemon: stop any running daemon, then relaunch it in detached mode. If the daemon is not running, it starts directly. `yeero restart [OPTIONS]` OptionDescription `-f, --force`Force stop the old process without waiting for graceful shutdown `# Graceful restart yeero restart # Force restart yeero restart --force` yeero login Log in with email verification code. OAuth is not supported in the CLI; use the web or desktop app for Google / GitHub login. `yeero login [OPTIONS]` OptionDescription `--email `Pre-fill email and skip interactive input `--code `Pre-fill verification code and skip sending step (requires `--email`) `# Interactive login yeero login # Non-interactive login (good for CI / scripts) yeero login --email you@example.com --code 123456` yeero logout Log out the current account. By default only this device is logged out and the local token is cleared; use `--all` to log out all devices. `yeero logout [OPTIONS]` OptionDescription `--all`Log the account out from all devices `yeero logout yeero logout --all` yeero whoami Print current account information in JSON format. Shows a hint when not logged in. `yeero whoami` yeero chat Start an AI conversation. Supports TUI full-screen mode, plain multi-turn mode and one-shot mode. See [Chat In Depth](https://yeero.ai/docs/en/cli-chat.html) for details. `yeero chat [MESSAGE] [OPTIONS]` Common options: | Option | Description | | --- | --- | | `MESSAGE` | Initial message (optional; omit to start an empty session) | | `--plain` | Use plain line mode (instead of TUI) | | `--oneshot` | Send a message and exit immediately; good for scripts / pipes | | `-m, --model ` | Model ID; repeat for multi-model parallel calls | | `--chat-id ` | Continue an existing conversation | | `--branch-id ` | Continue an existing branch | | `--folder-id ` | Folder for the new conversation | | `--tool ` | Enable a tool (repeatable) | | `--user-file ` | Attach a local file (repeatable) | | `--use-knowledge` | Enable knowledge base retrieval | `yeero chat "Explain vector databases" yeero chat -m gpt-4o -m claude-3-5-sonnet "Compare the two responses" yeero chat --oneshot "What is today's date?"` yeero chats List conversations and folders for the current account. `yeero chats [OPTIONS]` OptionDefaultDescription `--folder-id ``0` (root)Parent folder ID `--name `—Fuzzy search by name `--pinned`—Only return pinned items `--order ``DEFAULT``DEFAULT` or `EARLIEST` `--page ``1`Page number `--size ``50`Items per page (1–200) `--json`—Output as JSON (script-friendly) `yeero chats yeero chats --name "architecture" --json` yeero models List AI models available to the current account. `yeero models [OPTIONS]` OptionDescription `--keyword `Fuzzy search by name `--free`Show only free models `--input-modalities `Filter by input capabilities, comma-separated (e.g. `text,image`) `--output-modalities `Filter by output capabilities, comma-separated (e.g. `text,audio`) `--page / --size`Pagination options `--json`Output as JSON `yeero models yeero models --keyword claude --free yeero models --output-modalities image --json` yeero tools List tools available to the current account. `yeero tools [OPTIONS]` OptionDescription `--keyword `Fuzzy search by ID / name / description `--json`Output as JSON `yeero tools yeero tools --keyword search --json` yeero app Manage Python AI apps. Contains multiple subcommands; see [AI Apps](https://yeero.ai/docs/en/cli-app.html) for details. | Subcommand | Description | | --- | --- | | `yeero app create` | Create a new Python app | | `yeero app list` | List all apps | | `yeero app show` | Show app details | | `yeero app save` | Save local file as a new version | | `yeero app versions` | List version history | | `yeero app version-show` | Show details of a specific version | | `yeero app rollback` | Roll back to a specific version | | `yeero app generate` | AI code generation (single-turn) | | `yeero app discuss` | AI multi-turn discussion REPL | | `yeero app run` | Run an app | | `yeero app stop` | Stop a running app | | `yeero app uninstall` | Uninstall a locally installed app | ## yeero do Describe what you want in natural language, and yeero routes to and runs the most suitable Python AI app. See [AI App Execution](https://yeero.ai/docs/en/cli-do.html) for details. `yeero do [INTENT] [OPTIONS]` OptionDescription `INTENT`Natural-language intent. You will be prompted if omitted. `--dry-run`Preview the routing result without executing. `--name `Name for the newly created script when no app matches. `--no-create`Fail instead of creating a new app when no match is found. `--model `Model ID used to generate code when creating a new app. yeero config Manage local configuration. See [Config & Daemon](https://yeero.ai/docs/en/cli-config.html) for details. | Subcommand | Description | | --- | --- | | `yeero config get-lang` | Show current UI language | | `yeero config set-lang ` | Set UI language (`zh` or `en`) | | `yeero config get-default-model` | Show current default model | | `yeero config set-default-model ` | Set default model | | `yeero config unset-default-model` | Clear default model setting | | `yeero config add-to-path` | Add yeero directory to system PATH | | `yeero config get-path` | Show existing yeero entries in PATH | | `yeero config clear-path` | Remove all yeero entries from PATH | ## yeero update Check for and automatically install the latest CLI version. `yeero update [OPTIONS]` OptionDescription `--check`Only check for updates, do not download or install `# Check and install automatically yeero update # Check only, do not install yeero update --check` 💡 Auto update checkEvery time you run a CLI command, the CLI silently checks for updates in the background (cached for 1 hour). When a new version is available, it prints a single yellow hint without blocking the current command. --- # CLI Chat In Depth *Deep dive into `yeero chat`: plain multi-turn mode, one-shot mode, multi-model parallel, files & tools, knowledge base, media generation and sampling parameters.* > **⚡ TL;DR** > yeero chat supports multi-turn and one-shot modes, multi-model parallel replies, file and tool inputs, knowledge-base grounding and full sampling-parameter control. ## Two Interaction Modes ### Plain Multi-turn Mode (Default) Running `yeero chat` enters plain multi-turn mode by default: type your message after the terminal prompt, press Enter to send, and the AI streams its reply. You can continue the conversation for as many turns as you like; type `/exit` or press Ctrl+C to quit. `# Start plain multi-turn (new conversation) yeero chat # Start with an initial message yeero chat "Analyze this code snippet" # Resume an existing conversation yeero chat --chat-id 12345` One-shot Mode Add `--oneshot` to send a single message and exit immediately. Great for scripts, pipes and CI: `# Print the AI response directly yeero chat --oneshot "Summarize quantum computing in one sentence" # Pipe input cat error.log | yeero chat --plain --oneshot "Explain this error" # Disable color output (good for pipes or log collection) yeero chat --no-color --oneshot "Generate a JSON snippet"` Multi-model Parallel Pass `-m` multiple times to send the same message to several models at once and compare their responses side by side: `yeero chat -m gpt-4o -m claude-3-5-sonnet "Which programming language is better for AI development?" yeero chat -m deepseek-r1 -m gemini-2.5-pro "Solve this math problem: ..."` When no `-m` is given, the default model is used (set via `yeero config set-default-model`). ## Session & Branch Management | Option | Description | | --- | --- | | `--chat-id ` | Continue a specific conversation (from existing history) | | `--branch-id ` | Continue a specific branch | | `--folder-id ` | Folder for the new conversation | `# Find the conversation ID yeero chats --name "last project" # Continue that conversation yeero chat --chat-id 98765 "Continue the previous discussion"` Files & Tools File Attachments Use `--user-file` to upload local files as attachments (repeatable for multiple files): `yeero chat --user-file ./report.pdf "Summarize the key points of this report" yeero chat --user-file image1.png --user-file image2.png "Compare these two images"` Enable Tools Use `--tool` to specify a tool ID (run `yeero tools` to see available tools): `yeero tools # list available tools yeero chat --tool web_search "Any tech news today?"` Knowledge Base Retrieval `yeero chat --use-knowledge "Based on my past conversations, summarize my main technical interests"` Sampling Parameters Fine-tune model generation behavior: | Option | Range | Description | | --- | --- | --- | | `--temperature` | 0–2 | Sampling temperature; higher = more random (default depends on model) | | `--nucleus-sampling` | 0–2 | Top-P text diversity | | `--maximum-token-count` | positive integer | Maximum number of tokens to generate | | `--presence-penalty` | 0–2 | Topic diversity; reduces repeated topics | | `--frequency-penalty` | 0–2 | Vocabulary diversity; reduces repeated tokens | | `--reasoning-effort` | XHIGH｜HIGH｜MEDIUM｜LOW｜MINIMAL｜NONE | Reasoning depth (for models that support reasoning) | | `--message-window-size` | 0–100 | Context history message window size | `yeero chat --temperature 0.2 --reasoning-effort HIGH "Write a rigorous technical analysis" yeero chat --temperature 1.5 "Give me a creative story opening"` Image Generation Choose a model that supports image output (e.g. dall-e-3, gpt-image-1) and pass image parameters: | Option | Values | Description | | --- | --- | --- | | `--image-aspect-ratio` | RATIO_1_1｜RATIO_2_3｜RATIO_3_2｜RATIO_4_3｜RATIO_3_4｜RATIO_16_9｜RATIO_9_16 | Image aspect ratio | | `--image-size` | SIZE_1K｜SIZE_2K｜SIZE_4K | Image resolution | `yeero chat -m dall-e-3 --image-aspect-ratio RATIO_16_9 --image-size SIZE_2K "A cyberpunk cityscape at night"` Audio Generation Choose a model that supports audio output (e.g. gpt-4o-audio) and pass audio parameters: | Option | Values | Description | | --- | --- | --- | | `--audio-voice` | ALLOY｜ECHO｜FABLE｜ONYX｜NOVA｜SHIMMER | Synthesized voice | | `--audio-format` | WAV｜MP3｜FLAC｜OPUS｜PCM16 | Audio output format | `yeero chat -m gpt-4o-audio --audio-voice NOVA --audio-format MP3 "Read this text aloud: ..."` Video Generation Choose a model that supports video output and pass video parameters. Video generation is an async task; the CLI automatically polls for the result: | Option | Values | Description | | --- | --- | --- | | `--video-aspect-ratio` | RATIO_16_9｜RATIO_9_16｜RATIO_1_1｜RATIO_4_3｜RATIO_3_4 | Video aspect ratio | | `--video-resolution` | P480｜P720｜P1080｜K1｜K2｜K4 | Video resolution | | `--video-duration` | 1–300 (seconds) | Video duration | | `--video-generate-audio` | true｜false | Whether to generate an audio track | ``` yeero chat -m wan-2.1 --video-aspect-ratio RATIO_16_9 --video-resolution P720 --video-duration 10 "A cat running across a meadow" ``` --- # CLI AI Apps *Manage the full lifecycle of Python AI applications with `yeero app` — from creation and code generation to running, versioning, and rollback.* > **⚡ TL;DR** > yeero app manages the full lifecycle of Python AI apps from the terminal: create, generate code, run, version and roll back. ## Overview | Subcommand | Description | | --- | --- | | `app create` | Create a new app (cloud registration + local init) | | `app list` | List all apps and their status | | `app show` | Show app details | | `app save` | Save local file as a new version | | `app versions` | List version history | | `app version-show` | Show details of a specific version | | `app rollback` | Roll back to a previous version | | `app generate` | AI single-turn code generation | | `app discuss` | AI multi-turn REPL discussion | | `app run` | Run application locally | | `app stop` | Stop a running application | | `app uninstall` | Uninstall (remove script and venv) | ## Create an App `yeero app create --name "Data Analyst" --description "Analyze CSV data and generate reports"` AI Code Generation `# Single-turn generation yeero app generate my-app --prompt "Add data visualization with line charts" # Specify edit mode yeero app generate my-app --prompt "Optimize performance" --mode patch # Specify reasoning effort yeero app generate my-app --prompt "Refactor code structure" --effort high` OptionDefaultDescription `--prompt `interactiveGeneration instruction `--mode``auto`Edit mode: `auto` | `full` | `patch` `--effort``medium`Reasoning effort: `minimal` | `low` | `medium` | `high` | `xhigh` `--model `app configOverride model Multi-turn Discussion `yeero app discuss my-app yeero app discuss my-app --model claude-3-5-sonnet --effort high` Enters a multi-turn REPL. The AI reads the current code and iterates with you. Type `/exit` or press Ctrl+C to quit. ## Run & Stop `# Run an app yeero app run my-app # Run with initial input (JSON) yeero app run my-app --input '{"query": "sales last 30 days"}' # Stop a running app yeero app stop ` 💡 Auto dependency installOn first run, the CLI automatically creates an isolated Python virtual environment and installs required dependencies. ## Version Control ``` # Save current local file as new version yeero app save my-app --file ./app.py --change-summary "Added visualization" # List version history yeero app versions my-app # View a specific version yeero app version-show ``` ## Rollback ``` # Interactive selection (with confirmation prompt) yeero app rollback my-app # Roll back to a specific version yeero app rollback my-app --version -y ``` ## Uninstall ``` # Uninstall a single app yeero app uninstall my-app -y # Uninstall all locally installed apps yeero app uninstall --all -y ``` --- # AI App Execution *Describe what you want in one sentence — yeero matches or creates a Python app and runs it. No platform commands to memorize, no tokens burned on every run.* > **⚡ TL;DR** > yeero do turns one sentence into action: it matches or creates the right Python app and runs it locally — no per-run tokens, no commands to memorize. ## Overview `yeero do` is essentially still executing a Python app script from `yeero app`. The difference is that you do not need to remember an app ID or subcommand: just describe your intent in one sentence. The system will: 1. Search the locally synced Python apps for the best match; 2. Execute that app directly if the match is strong; 3. Create a new Python app generated by AI and then run it if no suitable app exists (or the match is weak). So `yeero do` works both for "I want to run an existing app" and for "I have a new idea; generate and run it for me." Another compelling scenario: **no more memorizing platform-specific commands**. Windows uses `ipconfig`, while Linux / macOS uses `ifconfig` or `ip addr`; checking processes, clearing caches, and managing services all have different commands across systems — now you can just type `yeero do "show my IP address"` without worrying about which OS you are on. More importantly, `yeero do` **dramatically cuts token costs**. Unlike AI agents that call an LLM on every interaction, `yeero do` only uses one LLM call when first creating the app; every subsequent run executes the local Python script directly — zero additional tokens consumed. For utility tasks you run frequently, switching from an agent to `yeero do` can save a significant amount of money. ## Syntax & Options `yeero do [INTENT] [OPTIONS]` Argument / OptionDescription `INTENT`Natural-language intent. You will be prompted if omitted. `--dry-run`Preview the routing result without actually executing. `--name `When no app matches, specify the name for the newly created script. Auto-generated by default. `--no-create`Fail instead of creating a new app when no match is found. `--model `Model ID used to generate code when creating a new app. Falls back to the app default if omitted. Execution Flow When you run `yeero do "..."`, the CLI first sends a dry-run probe: - **Strong match**: if one app clearly matches, it runs directly. If several apps score highly, a candidate list is shown for you to select with the up/down arrow keys (non-terminal environments fall back to typing a number). - **No match / weak match**: you are asked to confirm the model that will generate the new app, then the app is created, installed, and executed. During execution, stage messages such as `creating`, `generating`, `installing`, and `running` are printed to the terminal in real time. ## Examples ``` # I want something done — no need to know how; yeero matches or creates an app yeero do "sort these 200 photos into folders by date taken" # No need to memorize cross-platform commands — no looking up Windows vs Linux syntax yeero do "show which process is using the most memory" # Daily automation — one sentence for repetitive work yeero do "deduplicate this CSV, sort by amount, and export to Excel" # Preview which app would match, without actually running it yeero do "generate a Pomodoro timer web page" --dry-run # Specify a name and model when creating a new app yeero do "write a script to back up the database and upload to S3 daily" --name db-backup --model gpt-4o # Fail on no-match instead of creating a new app yeero do "just testing" --no-create ``` ## Difference from yeero app run | | `yeero do` | `yeero app run` | | --- | --- | --- | | Input | Natural-language intent | App ID / name | | Matching | Automatic routing; creates a new app if needed | Runs the specified app directly | | Best for | “I want to do X” | “I want to run this app” | | Essence | Both ultimately execute a Python app script from `yeero app` | ## Notes - `yeero do` requires the daemon; it will be started automatically on demand. - The first time an app is created, you may wait for AI code generation and dependency installation. - If the generated app fails, the CLI prints the app ID so you can continue iterating with `yeero app discuss `. --- # Config & Daemon *Use `yeero config` to manage language, default model, and PATH settings. Use `yeero start / stop / status` to control the background Daemon service.* > **⚡ TL;DR** > yeero config manages language, default model and PATH settings, while yeero start / stop / status controls the background daemon that powers the CLI. ## Config Subcommands `yeero config` provides the following subcommands, all of which are pure local file operations — no Daemon required: | Subcommand | Description | | --- | --- | | `config get-lang` | Show the current UI language | | `config set-lang` | Set the UI language | | `config get-default-model` | Show the current default model | | `config set-default-model` | Set the default model | | `config unset-default-model` | Clear the default model (restore built-in default) | | `config add-to-path` | Add yeero to the system PATH | | `config get-path` | Show yeero entries already in PATH | | `config clear-path` | Remove all yeero entries from PATH | ## Language Settings yeero CLI supports two UI languages: Chinese (`zh`) and English (`en`). `# Show current language yeero config get-lang # Set to English yeero config set-lang en # Set to Chinese yeero config set-lang zh # Run without argument for interactive input yeero config set-lang` The language setting is persisted in `config.toml`, isolated by build environment (dev / pre / prod). > **💡 Global --lang flag** > Any command also accepts `--lang en` / `--lang zh` to temporarily override the language, taking priority over the config file. ## Default Model The default model is used by `yeero chat`, `yeero app generate`, and other commands when `--model` is not specified. `# Show current default model yeero config get-default-model # Set a default model yeero config set-default-model gpt-4o # Run without argument for interactive input yeero config set-default-model # Clear the default model (revert to built-in default) yeero config unset-default-model` You can get the full list of available model IDs with `yeero models`. ## Add yeero to PATH If your terminal cannot find the `yeero` command after installation, run the following to configure PATH automatically: `# Add to user-level PATH (recommended) yeero config add-to-path # Add to system-level PATH (requires admin/sudo) yeero config add-to-path --system # Preview without making changes yeero config add-to-path --dry-run # Remove from PATH yeero config add-to-path --remove` OptionDescription `--system`Write to system-level PATH (default: user-level) `--remove`Remove the yeero directory from PATH `--dry-run`Preview only, no actual changes 💡 macOS / LinuxThis command appends a PATH export line to `~/.bashrc`, `~/.zshrc`, or similar shell config files. Reopen your terminal or run `source ~/.zshrc` for the change to take effect. ## Check PATH Configuration Use `config get-path` to view the current yeero executable directory and any yeero entries present in both user-level and system-level PATH: `yeero config get-path` Example output: `current yeero directory: /usr/local/bin user PATH yeero entries: /usr/local/bin system PATH: no yeero entries found` Remove yeero from PATH `config clear-path` removes all yeero-related entries from PATH (without appending a new path). Useful for uninstalling or relocating: `# Remove from user-level PATH (default) yeero config clear-path # Remove from system-level PATH (requires admin/sudo) yeero config clear-path --system # Preview without making changes yeero config clear-path --dry-run` OptionDescription `--system`Remove from system-level PATH (default: user-level) `--dry-run`Preview only, no actual changes 💡 Relationship to add-to-path --remove`clear-path` has the same effect as `add-to-path --remove` — it is a more semantic alias designed for intuitive use. ## Daemon Overview Most yeero CLI features (chat, AI apps, etc.) rely on a background Daemon process. The Daemon runs as a local HTTP service, handling communication with cloud APIs, managing conversation context, and controlling AI app lifecycles. Under normal circumstances, the Daemon starts automatically when you run any network-requiring command. You can also manage it manually: ## Start the Daemon `# Start Daemon in background yeero start # Specify a listening port (default: auto-assigned) yeero start --port 7700 # Run in foreground (for debugging) yeero start --foreground` Stop the Daemon `yeero stop` This command sends a termination signal to the running Daemon for a graceful shutdown. ## Check Status `yeero status` Example output: `Daemon running on port 7700 Version: 0.1.0` If the Daemon is not running, it will display `Daemon not running`. ## Config File Location yeero's config file `config.toml` is stored per build environment: | OS | Path | | --- | --- | | macOS / Linux | `~/.config/yeero/config.toml` | | Windows | `%APPDATA%\yeero\config.toml` | The config file is in TOML format and currently supports the following fields: `# UI language: zh or en lang = "en" # Default model ID (optional) default_model = "gpt-4o"` ⚠ NoteDo not manually edit other fields in config.toml. Always use `yeero config` commands to modify settings. --- # Desktop *Yeero Desktop is a native desktop application built with Tauri, delivering the full YeeroAI experience with native-level performance and system integration.* > **⚡ TL;DR** > Yeero Desktop is a Tauri-based native app for Windows, macOS and Linux — the full YeeroAI experience with native performance, global Launcher, capture tools and built-in editors. ## Overview Yeero Desktop packages the full YeeroAI feature set as a native installer for macOS, Windows, and Linux. Compared to the web app, the desktop client additionally provides: - Launch directly from the system Dock / taskbar without opening a browser - Native window title bar with automatic light/dark theme following the system setting - In-app auto-updates — install new versions with a single click - OS system notifications (including update alerts) ## Installation Download the installer for your operating system and architecture: | OS | Architecture | Download | | --- | --- | --- | | macOS | Apple Silicon (M-series) | [Yeero-latest-aarch64.dmg](https://yeero.ai/desktop/latest/Yeero-latest-aarch64.dmg) | | macOS | Intel | [Yeero-latest-x64.dmg](https://yeero.ai/desktop/latest/Yeero-latest-x64.dmg) | | Windows | x64 | [Yeero-latest-x64.exe](https://yeero.ai/desktop/latest/Yeero-latest-x64.exe) | | Windows | ARM64 | [Yeero-latest-arm64.exe](https://yeero.ai/desktop/latest/Yeero-latest-arm64.exe) | | Linux | x64 | [Yeero-latest-amd64.AppImage](https://yeero.ai/desktop/latest/Yeero-latest-amd64.AppImage) | | Linux | x64 (deb) | [Yeero-latest-amd64.deb](https://yeero.ai/desktop/latest/Yeero-latest-amd64.deb) | ### macOS 1. Download the `.dmg` file for your chip 2. Open the DMG and drag **Yeero** to the Applications folder 3. On first launch, if you see "cannot verify developer", right-click Yeero.app → **Open** and confirm in the dialog; or go to **System Settings → Privacy & Security** and click **Open Anyway** > **💡 How to check your Mac chip** > Click the Apple menu → **About This Mac**. A chip starting with "M" means Apple Silicon; "Intel" means x64. > **⚠ "Yeero is damaged and can't be opened"** > Because the app is not yet notarized with an Apple Developer certificate, macOS may show this error. To fix it, open Terminal and run: `xattr -cr /Applications/Yeero.app` Then try opening the app again. ### Windows 1. Download the `.exe` installer 2. Run the installer; if a SmartScreen warning appears, click **More info** → **Run anyway** 3. Follow the setup wizard to complete installation. Yeero will appear in the Start Menu and as a desktop shortcut ### Linux **AppImage (recommended, no installation required):** `# Make executable chmod +x Yeero-latest-amd64.AppImage # Run directly ./Yeero-latest-amd64.AppImage` **deb package:** `sudo dpkg -i Yeero-latest-amd64.deb` ⚠ Linux dependenciesThe Linux build requires `libwebkit2gtk-4.1` and other system libraries. If Yeero fails to start, run: `sudo apt-get install libwebkit2gtk-4.1-0 libgtk-3-0 libayatana-appindicator3-1` ## First Launch When you open Yeero Desktop for the first time: 1. The app automatically starts the built-in Daemon service (no manual setup needed) 2. The main interface loads once the Daemon is ready 3. Click the avatar or **Sign In** button in the top-right corner to log in 4. After signing in, all features are available ## Features Yeero Desktop includes all web features plus desktop-exclusive capabilities: - **All Web Features** — multi-model chat, branching thinking, knowledge base, AI apps, 20+ built-in tools, file library, folders, prompt templates and more - **Launcher** — Alt+Space global command bar: calculations, app launching, file search, screenshots, recording, clipboard history and AI chat - **Screenshot & Recording** — smart window detection, magnifier with RGB picker, long screenshots with auto-scroll stitching, system audio and microphone, camera picture-in-picture - **Image Editor** — standalone window with 9 annotation tools, layers panel, EXIF metadata viewer - **Video Editor** — multi-track timeline editing, 7 transitions, filters and color grading, text and subtitles, AI-powered Smart Trim - **Disk Cleaner** — visualize disk usage with a directory tree, auto-identify apps across 30+ categories, AI-analyze unknown folders ## Theme Yeero Desktop supports three theme modes, switchable from the Settings page or the top-right corner of the interface: - **Light** — always use the light theme - **Dark** — always use the dark theme - **System** — automatically follow the macOS / Windows system theme Switching themes also updates the native title bar color for a consistent visual experience. ## Language You can switch the app UI between English and Chinese from the Settings page. The language setting is independent from the CLI and does not affect it. ## Auto-Update Yeero Desktop has a built-in auto-update mechanism: - Checks for new versions automatically on each launch (at most once per hour) - Also checks when the window is brought back to the foreground - When a new version is found, an in-app prompt appears and a system notification is sent - Click **Update Now** in the prompt — the update downloads and installs automatically on restart All update packages are verified with Ed25519 digital signatures to ensure security. > **💡 Manual update check** > You can also click **Check for Updates** on the Settings page to trigger a check immediately, bypassing the 1-hour cache. ## Built-in Daemon & CLI Yeero Desktop bundles the full `yeero` CLI and runs the Daemon as a sidecar process alongside the app. The Daemon starts automatically when the app launches and shuts down when the app exits — no manual management required. **Installing Desktop gives you the yeero command in your terminal.** Desktop and the CLI share the same Daemon backend, which means: - After signing in through Desktop, open a terminal and run `yeero chat` directly — no separate login needed - All CLI conversations, configuration, and knowledge base data are fully synced with Desktop - No need to install the CLI separately — the Desktop installer already includes the complete `yeero` binary > **💡 Using the yeero command in terminal** > After installing Desktop, the `yeero` command is automatically added to your system PATH (macOS / Linux) or becomes available via system environment variables (Windows). Open a new terminal window and run `yeero --help` to verify. --- # Launcher *The Launcher is Yeero Desktop's global command bar: press `Alt+Space` and a single input box handles calculations, app launching, file search, screenshots, screen recording, clipboard history, and even direct AI questions — without leaving your current work.* > **⚡ TL;DR** > The Launcher is a global Alt+Space command bar: calculations, app launching, file search, screenshots, screen recording, clipboard history and direct AI questions — without leaving your current work. > **💡 Desktop only** > The Launcher relies on a system-wide global hotkey and local capabilities (app enumeration, window control, file indexing…), so it is only available in Yeero Desktop, not on the web. ## Overview The Launcher is built around one idea: "one input box, everything solved". Anything you type is parsed by 16 processors in parallel — expressions go to the calculator, URLs to the browser, app names to app search, natural language to AI. All results are merged, ranked by relevance, and executed with a single Enter. - **Global hotkey** — `Alt+Space` by default, works from any application; press again to hide - **Bilingual triggers** — every processor accepts both Chinese and English keywords; "截图" and "screenshot" do the same thing - **Pinyin & typo tolerance** — full pinyin, pinyin initials, homophones and misspelling correction are all supported - **Two presentation modes** — most results execute immediately on selection (e.g. launching an app); some capabilities (system monitor, settings, HTML apps) open a panel inside the Launcher window and keep it open - **Idle quick panel** — with an empty query, the Launcher shows frequently used actions (lock, sleep, mute…) one click away ## Quick Start 1. Press `Alt+Space` to summon the Launcher 2. Just type: try `12*8`, `screenshot`, `chrome`, or any question 3. Navigate with `↑` `↓`, or press `Enter` to run the top result After you stop typing, the text is dispatched to every enabled processor, and results are listed from best to worst match. Each result shows its source processor and a hint; hover to see available shortcuts. ## Matching The matching engine is deeply optimized for mixed Chinese/English input: - **Fuzzy matching** — no need to remember full names; typing a few scattered letters still hits (e.g. `ysbjsq` finds 记事本) - **Pinyin support** — match Chinese names with full pinyin or initials - **Homophones** — homophone variants still match - **Typo correction** — a few wrong or missing letters don't get in the way; you still get results - **Multi-field scoring** — app search weighs display name, aliases, localized names, descriptions and more, so the most likely intent ranks first ## The 16 Processors at a Glance The table summarizes all built-in processors. The trigger column shows typical examples — more synonyms are accepted. | Processor | Example triggers | What it does | | --- | --- | --- | | Calculator | `12*(3+5)` | Live expression evaluation, copy result | | URL Opener | `google.com` | Open URLs in your default browser | | Screenshot | `screenshot` | Full-screen region capture, then image editor | | Screen Record | `screen record` | Region/window recording with a floating control bar | | System Commands | `lock` / `锁屏` | 20+ system commands with confirmation for risky ones | | System Monitor | `sm` / `cpu` | Live CPU/memory/disk/network panel | | Yeero Navigation | `settings` / `knowledge` | Jump to app pages, lifecycle actions, open editors | | App Search | any app name | Launch installed apps, pinyin/typo tolerant | | Bookmarks | `bm github` | Search bookmarks from 4 major browsers | | Clipboard History | `clip @image` | Search and restore clipboard history with rich filters | | Kill Process | `kill chrome` | Terminate processes by name (grouped by name) | | HTML App | any app name | Run BROWSER-type AI apps in a Launcher panel | | File Search | any file name (2+ chars) | Blazing-fast local index search, optional deep content search | | Window Switcher | any window title | Focus existing windows, including minimized ones | | Routing Engine | any natural language | AI intent routing to run the right AI app | | AI Chat | any question | Ask your default model directly, streaming answers | ## Processor Reference ### Calculator Type an expression made of digits and operators and the result appears live in the candidate list — no need to press Enter to get the answer. Supported: `+` `-` `*` `/` `^` (power), `%` (modulo), parentheses, negatives and decimals. - **Live results** — the answer is shown as the first candidate while you type - **Enter to copy** — press `Enter` to copy the result to the clipboard, ready to paste into documents, spreadsheets or chats - **Clean output** — `0.1+0.2` yields a clean `0.3`, with no floating-point tail - **Idle quick entry** — the idle panel also offers a calculator shortcut when nothing is typed Examples: `12*(3+5)` → `96`; `2^10` → `1024`; `100%7` → `2`; `-8/4` → `-2`. ### URL Opener Type a full link (`https://…`) or just a bare domain like `github.com` or `yeero.ai` — Enter opens it in your default browser. - **Auto protocol** — bare domains are automatically completed with an `https://` prefix - **Ports & paths** — addresses with ports or intranet hosts like `localhost:3000/admin` or `192.168.1.1` work too ### Screenshot Type "screenshot" (or 截图/截屏; the first few letters are enough to suggest it) and press Enter. The Launcher hides itself and enters full-screen region selection — the whole flow takes three steps: 1. **Select the area** — drag to select any region, or simply click a window to let the system detect its bounds automatically 2. **Confirm** — a standalone image editor window opens automatically 3. **Annotate & export** — highlight, draw and add text right away, then save to your file library or copy to the clipboard The idle panel also offers a screenshot shortcut when nothing is typed. For the full capture workflow (smart window detection, long-screenshot stitching) and annotation tools, see [Screenshot & Recording](https://yeero.ai/docs/en/capture.html). ### Screen Record Type "screen record", "screenrecord", 录屏 or 屏幕录制 (spaces are optional) and press Enter: 1. **Choose the recording area** — drag a region (minimum 100×100), or double-click a window for automatic detection 2. **Full control while recording** — a floating control bar appears on screen with a status indicator and live duration, supporting pause, resume and stop 3. **Ready when you finish** — after recording, send the video straight to the video editor, or view the file in your file manager - **No interruptions** — while a recording is in progress, the Launcher stops showing recording candidates; all controls live on the floating bar - **Adjustable quality** — frame rate and quality are configurable in [Settings → Launcher](https://yeero.ai/docs/en/settings.html) - **Idle quick entry** — the idle panel offers a recording shortcut, no typing needed Recording also supports microphone audio and camera picture-in-picture — see [Screenshot & Recording](https://yeero.ai/docs/en/capture.html) for details. ### System Commands Common system operations as searchable commands — one keyword gets you there, no more digging through system menus. Both Chinese and English triggers work, English hyphens are optional (`volumeup` and `volume-up` both work), and typing the first few letters suggests matches live: | Command | English trigger | Chinese trigger | | --- | --- | --- | | Lock screen | lock | 锁屏 | | Sleep | sleep | 睡眠 | | Shut down | shutdown | 关机 | | Restart | restart | 重启 | | Turn off display | turn-off-display | 关闭屏幕 / 息屏 | | Mute | mute | 静音 | | Volume up / down | volume-up / volume-down | 音量加 / 音量减 | | Media play/pause | media-play-pause | 播放暂停 / 播放 / 暂停 | | Previous / next track | media-previous / media-next | 上一首 / 下一首 | | Emoji panel | emoji-panel | 表情 / 表情符号 | | Task manager | task-manager | 任务管理器 | | System settings | system-settings | 系统设置 / 设置 | | Recycle bin / empty it | recycle-bin / empty-recycle-bin | 回收站 / 清空回收站 | | Show desktop | show-desktop | 显示桌面 | | Open Downloads / Documents | open-downloads / open-documents | 打开下载 / 打开文档 | | Home directory | open-home | 用户目录 / 主目录 / 家目录 | | Recent files | recent / recent-files | 最近文件 / 最近 | Every command accepts further synonyms (e.g. "音量增", "音量降", "下载", "文档" in Chinese) — type it whichever way feels natural. - **Protection for risky actions** — shutdown, restart and emptying the recycle bin always ask for confirmation first, so a slip of the hand is safe - **Recent files at hand** — type "recent files" (or `recent`) to list the 20 most recently opened files; Enter opens them with their associated application - **Always in the idle panel** — with an empty query, the idle panel always offers seven high-frequency commands: lock, sleep, show desktop, turn off display, mute, system settings and task manager — one click away ### System Monitor Type `sm`, `monitor`, 系统监控, 监控 or 系统状态 and a live system monitor panel expands inside the Launcher window, continuously refreshed — no need to open the task manager. - **Complete coverage** — CPU usage and core info, memory and swap usage, capacity and usage of every disk, network upload/download speeds, sensor temperatures, system load, process count, OS version and uptime, all on one screen - **Sub-keyword jump** — type "CPU", "memory" or "disk" (also `disk`, 硬盘, 存储) and the panel opens scrolled to the matching section - **Panel stays in the window** — the panel renders inside the Launcher window; press `Esc` to close it and get back to work - **Idle quick entries** — the idle panel offers shortcuts for System Monitor, CPU, Memory and Disks ### Yeero Navigation Navigate and act inside Yeero with natural language — all entries accept Chinese and English, and you may prepend verbs like "open", "go to" or 打开/前往/跳转 ("open settings" works exactly like "settings"). Matches appear in three ways: standalone pages open in a new window, while settings and account panels expand right inside the Launcher window without interrupting your work. - **Open pages (new window)** — "chats", "new chat", "knowledge", "folders", "files" (file library), "models"; saying "message folders" / "app folders" opens the folders page directly on the matching category - **Settings, down to the category** — "settings" opens the full settings panel; "system settings", "general settings", "prompt settings", "memory settings", "launcher settings" and "file index settings" jump straight to their category, rendered inside the Launcher window - **Account, down to the tab** — "account" / "profile" opens the account panel; "credits", "invitation" and "feedback" jump straight to their tab - **Lifecycle actions** — "main window" / "home" focuses the main window, "quit" exits the app, "restart" relaunches it - **Open editors** — "image editor", "photo editor", "video editor" and "剪辑" (edit) open their standalone windows directly - **Help docs** — "docs" / "help" opens this documentation site in your browser, in your current language - **Media files direct** — type or paste an image/video file path: images open in the image editor, videos in the video editor. Quoted paths and `file://` links are supported, and "open/edit" verbs before or after the path are recognized too (e.g. `open C:\photos\avatar.png`); even a pasted image goes straight into the image editor ### App Search Anything you type also searches your installed applications — much faster than digging through the Start menu or Launchpad. Candidates show real app icons; Enter launches the app. - **Type it any way** — app names, full pinyin (e.g. `weixin`), pinyin initials, even keywords from an app's description all match - **Chinese & English interchangeable** — type "微信" to find WeChat, or type "wechat" to find 微信 - **Homophones & typo correction** — homophone variants and small misspellings still produce results - **Always up to date** — after installing or uninstalling apps, results update automatically with nothing to do ### Bookmarks Type the prefix `bm`, `bookmark` or 书签 followed by keywords to search bookmarks across Chrome, Edge, Firefox and Safari at once; Enter opens one in your default browser. - **Smart ranking** — exact title matches come first, followed by title-prefix, title-contains and URL-contains matches, so what you want is always near the top - **Complete info** — every bookmark shows its title, folder and URL at a glance - **Browse without keywords** — with just the prefix, the first 50 bookmarks are listed for direct browsing - **Idle quick entry** — the idle panel offers a bookmarks shortcut, no typing needed ### Clipboard History Type the prefix `clip`, `clipboard` or 剪贴板 to search everything you have ever copied — text, images and files are all captured automatically. Select a history entry and press Enter to restore it to the clipboard, ready to paste; never lose freshly copied content to an accidental overwrite again. - **Keyword search** — type any keywords after the prefix to fuzzy-search the content - **Filter syntax** — the filters below combine freely with keywords, in any order: | Syntax | Meaning | | --- | --- | | `@text` (`@t` / `@txt`) | Text entries only | | `@image` (`@i` / `@img`) | Image entries only (with thumbnail preview) | | `@file` (`@f`) | File entries only | | `@tag:work` | Filter by tag | | `@today` (`@td`) / `@week` / `@month` / `@year` | Today / this week / this month / this year | | `@7d` | Last N days (any number) | | `@date:2026-08-01~2026-08-27` | Explicit date range | For example, `clip @image @week design` means "images containing 'design' copied this week". - **Rich entry info** — every entry shows its type, copy time, sync status and pinned/favorite marks; image entries render thumbnail previews directly - **Cross-device sync, end-to-end encrypted** — with cloud sync enabled, your clipboard history is shared automatically across devices signed in to the same account; synced content is end-to-end encrypted with keys stored only on your own devices, unreadable to anyone else. Encrypted entries synced from other devices show a hint when they cannot be decrypted here yet, and become viewable once you configure the same key in [Settings → System](https://yeero.ai/docs/en/settings.html) - **Idle quick entry** — the idle panel offers a clipboard history shortcut to browse your recent copies without typing History limit, recording toggle and cross-device cloud sync live in [Settings → Launcher](https://yeero.ai/docs/en/settings.html); the end-to-end encryption key is configured in [Settings → System](https://yeero.ai/docs/en/settings.html). ### Kill Process When an app freezes or stops responding, type the prefix `kill` (or 结束/终止/杀掉) followed by the process name to end it quickly — no need to dig through the task manager. - **Search in Chinese or pinyin** — "微信" matches the Weixin process; `xunlei` matches 迅雷; small spelling deviations still hit - **Auto-grouped by name** — one app often spawns many background processes; results are merged into a single group showing the instance count and total memory, and Enter ends the whole group at once - **Readable names** — friendly localized names and icons are shown where available ("迅雷" instead of Thunder.exe), so you never face raw process file names - **Clear failures** — insufficient permissions or already-exited processes are reported explicitly, never silently ### HTML App Your [AI Apps](https://yeero.ai/docs/en/ai-apps.html) (BROWSER type) run right inside the Launcher: type an app name (pinyin and typo correction supported) and selecting the match opens it in a panel within the Launcher window — no need to switch to the main window; use it and move on. - **Always fresh** — every time you summon the Launcher, the app list refreshes automatically, so newly created apps are available immediately - **Full experience in the panel** — the app runs completely inside the panel while the Launcher stays open for continuous use - **Graceful fallback** — when nothing matches, a "Create in Apps" entry takes you to the app creation flow with your input prefilled ### File Search Any query of 2+ characters participates in file search. Results come from Yeero's own file index — much faster than system search, and the files you open most often rank higher over time. Scan scope and ignore rules are configured in [Settings → File Index](https://yeero.ai/docs/en/settings.html): - **First-time wizard** — your first search walks you through index setup; pick the scan scope and you're ready - **Search while indexing** — while the index is being built, a scanning progress indicator is shown and already-indexed files stay searchable - **Complete result info** — every result shows the file name (with keyword highlighting), location, size and modified time; Enter opens it with its associated program - **Deep content search** — add the `deep:` prefix (e.g. `deep:quarterly report`) to search file *contents* via the OS native index (Spotlight / Windows Search / Tracker); off by default and enabled in settings, with a guidance hint when disabled - **Helpful empty states** — empty results show how many files are currently indexed and offer an entry to adjust the scan scope, so you can tell whether the index simply doesn't cover your files - **Quick actions** — results support `Alt+Enter` (reveal in file manager) and `Ctrl/Cmd+Enter` (copy full path) ### Window Switcher Any text matches against titles, app names and process names of open windows; Enter restores the window and brings it to the front — instead of opening a duplicate, any window is one jump away no matter how many are open. - **Every window counts** — including minimized and tray-hidden ones, with their current state labeled on each entry - **Just as smart matching** — pinyin, homophones and typo correction all work; Chinese window titles can be found via pinyin too - **Always a fresh list** — windows are re-scanned on every Launcher invocation; if the target window was just closed, its entry disappears automatically so you never click a dead window - **Permission guidance** — macOS asks for automation permission on first use, with a one-click jump to System Settings; unsupported Linux environments (e.g. Wayland) explain the reason and remedy clearly ### Routing Engine The routing engine makes "run an app with one sentence" possible: just describe what you want in plain words (e.g. "translate this text into English", "sum the sales in this spreadsheet") — the AI understands your intent and finds the AI app that gets it done. - **Match and run** — when an app matches, it appears directly as a candidate; Enter runs it, with the execution process and results shown live in the Launcher panel - **Iterate in the panel** — after a run, type new input in the panel and press Enter to re-run the same app — no starting over - **Always a way forward** — when no existing app fits, two entries are offered: **"Quick create & run"** has the AI generate a new app on the spot and run it immediately; **"Create in Apps"** opens the app creation page with your input prefilled — ideal when you want to clarify requirements first To learn more about the app system, see [AI Apps](https://yeero.ai/docs/en/ai-apps.html). ### AI Chat Any non-empty input produces an "Ask AI" candidate — selecting it starts a conversation with your default model right in a Launcher panel, with word-by-word streaming output that you can stop at any time. - **Specific input ranks higher** — the more complete and specific your question, the higher the candidate ranks, so you can often just press Enter to ask - **Follow up in the panel** — after the answer completes, type a new question in the panel and press Enter to continue with full context - **Built for quick answers** — look up a word, ask a concept, grab a code snippet without opening the main window; for multi-model comparison, branching or full conversation management, use [Chat](https://yeero.ai/docs/en/chat.html) in the main window ## Keyboard Shortcuts | Shortcut | Action | | --- | --- | | `Alt+Space` | Summon / hide the Launcher globally | | `↑` / `↓` | Move between candidates | | `Enter` | Execute selection; during panel execution, acts as iterative input to re-run | | `Ctrl+1` – `Ctrl+9` | Execute the Nth result directly | | `Alt+Enter` | Reveal in file manager (file search results) | | `Ctrl/Cmd+Enter` | Copy file path (file search results) | | `Alt+P` | Pin / unpin the selected result | | `Alt+M` | Open the context menu of the selection | | `Tab` | Toggle the details panel | | `Esc` | Close the details panel first, then the Launcher; cancels a running action | ## Related Settings Available under [Settings → Launcher](https://yeero.ai/docs/en/settings.html) (this category only appears on desktop): - **Summon hotkey** — shows the current global hotkey (default `Alt+Space`) - **Screen record FPS** — 5–30, default 25; higher means smoother but larger files - **Screen record quality** — H.264 CRF value 0–51, default 23; lower means better quality - **Clipboard history limit** — 10–10,000 entries - **Clipboard history** — when off, clipboard content is no longer captured - **Clipboard cross-device sync** — sync clipboard history across devices via the cloud, with "Sync Now" and "Clear Cloud Data" actions - **Processor toggles** — enable or disable any of the 16 processors individually, searchable by name and keywords; a disabled processor stops producing results > **💡 Processor toggles apply instantly** > Toggles are stored locally and take effect immediately — no restart needed. If a category of results seems to have "disappeared", check here first. ## Related Reading - [Screenshot & Recording](https://yeero.ai/docs/en/capture.html) — regions, annotation, long screenshots and recording controls - [Image Editor](https://yeero.ai/docs/en/image-editor.html) — annotation, layers and metadata tools after capture - [Video Editor](https://yeero.ai/docs/en/video-editor.html) — editing, transitions and export after recording - [AI Apps](https://yeero.ai/docs/en/ai-apps.html) — the app system behind HTML App and the Routing Engine --- # Screenshot & Recording *The complete screen-capture toolkit in Yeero Desktop: screenshots with smart window detection, an RGB-picking magnifier and long-screenshot stitching by scrolling; recording with region / fullscreen / window detection, system sound and microphone, camera picture-in-picture — all flowing seamlessly into the image and video editors.* > **⚡ TL;DR** > Yeero Desktop captures anything: screenshots with smart window detection, an RGB-picker magnifier and long-screenshot stitching; recording with system sound, microphone and camera PiP — everything flows into the built-in editors. > **💡 Desktop only** > Screenshot and recording rely on system-level screen capture, so they are only available in Yeero Desktop. Start either one by typing "screenshot" or "screen record" in the [Launcher](https://yeero.ai/docs/en/launcher.html). ## Screenshot ### Choosing a Region When you start a screenshot, the screen enters a full-screen selection overlay with two ways to define the area: - **Drag a region** — drag out any rectangle; the live dimensions (e.g. 1920 × 1080) are shown beside the selection - **Click a window** — click any window and the system detects its bounds automatically, no manual alignment needed A magnifier follows the cursor for pixel-precise alignment and shows the current pixel's **RGB value** in real time — color picking and alignment in one step. ### Actions After Selecting Once the region is set, an action bar appears: - **Copy** — the screenshot goes straight to the clipboard, ready to paste - **Edit** — open the [image editor](https://yeero.ai/docs/en/image-editor.html) to annotate, then save or share - **Long screenshot** — see the next section - **Cancel** — discard the selection and start over ### Long Screenshots Web pages and chat histories that exceed one screen can be captured in full with long screenshots. Two capture modes: - **Manual scroll** — scroll the target window with your mouse wheel or touchpad as usual; frames are captured and stitched automatically in the background - **Auto scroll** — click "Start auto scroll" and do nothing else: the system scrolls and captures on its own (keep your hands off the mouse while it runs), and you can stop anytime Live progress shows the frame count and accumulated height; capture stops automatically at the maximum height, and the stitched result opens in the editor as one tall image. ## Screen Recording ### Recording Settings Starting a recording first opens the settings panel: - **Frame rate** — the adjustable range and default are configured in [Settings → Launcher](https://yeero.ai/docs/en/settings.html); higher is smoother but produces larger files - **Quality** — High / Medium / Low (mapping to different encoding quality parameters) - **Format & output directory** — choose where recordings are saved ### Choosing the Recording Area - **Full screen** — record the entire display - **Drag a region** — record any rectangle, minimum 100×100 (smaller regions show a "region too small" warning) - **Double-click a window** — the window's bounds are detected automatically and used as the recording area ### Controls While Recording While recording, a floating control bar stays on screen showing the elapsed duration, and you can: - **Pause / resume** — take a break without breaking the recording; no gap on the timeline - **Stop** — finish the recording and open the completion panel ### Audio & Camera - **System audio** — capture what the computer plays (meetings, videos, notification sounds) - **Microphone** — capture your voice-over, with input devices other than the default selectable - **Camera** — overlay your webcam as picture-in-picture with an adjustable position, ideal for on-camera tutorials and demos ### When Recording Finishes After stopping, a completion panel lists the video segments produced, and you can: - **Open in Editor** — jump straight into the [video editor](https://yeero.ai/docs/en/video-editor.html) - **Open folder / Copy path / Download** — find the recording file quickly - **Export segments** — export segments as finished files Two reliability notes: video processing tools (ffmpeg) are downloaded automatically on the first recording — a one-time wait; and if a previous recording was interrupted unexpectedly, the next recording start automatically attempts a repair and reports the outcome. ## Related - [Image Editor](https://yeero.ai/docs/en/image-editor.html) — annotate and share screenshots - [Video Editor](https://yeero.ai/docs/en/video-editor.html) — edit and export recordings - [Launcher](https://yeero.ai/docs/en/launcher.html) — the fastest way to start screenshots and recordings --- # Image Editor *A lightweight image editor built into Yeero Desktop: jump straight in after a screenshot, or open any local image. Annotate with 9 tools — rectangles, arrows, mosaic and more — then crop, rotate, resize and knock out the background with smart transparency, all backed by layer management and an EXIF viewer, covering everyday touch-ups and bug-report screenshots.* > **⚡ TL;DR** > The built-in image editor covers everyday annotation: rectangles, arrows, mosaic and 6 more tools, with layer management and an EXIF viewer — and screenshots open straight into it. ## Overview The image editor runs in its own window, independent of the main window, and multiple instances can be open at once. It is designed around "annotate fast, share faster": - Supports PNG, JPG, JPEG, WebP, BMP and GIF - 9 annotation tools, each with a number hotkey — select and go - Image processing: crop, rotate, flip, resize and smart background transparency - Full undo/redo history; every action can be reverted - A layers panel to manage all annotations: visibility, ordering, deletion - One-click copy to clipboard, paste straight into chat apps or documents - Built-in image info panel with basic parameters and full EXIF data ## Opening Images There are four ways to get into the editor: 1. **Automatically after a screenshot** — once you confirm the region via [Screenshot](https://yeero.ai/docs/en/capture.html) or the [Launcher](https://yeero.ai/docs/en/launcher.html), the editor window opens with the capture loaded 2. **Open a file via the Launcher** — paste or type an image path in the Launcher and press Enter to open it in the editor 3. **Open from the menu** — press `Ctrl+O` or click the "Open Image" button to pick a local file 4. **Drag & paste** — with an empty window, drag an image file in, or press `Ctrl+V` to paste an image from the clipboard The window automatically resizes itself to fit the actual image dimensions, so small images don't stretch across the screen and large images are never cropped. ## Annotation Tools The toolbar sits at the top of the window; every tool has its own hotkey (on narrow windows some tools are folded into a "More tools" dropdown): | Hotkey | Tool | Description | | --- | --- | --- | | `V` | Select | Select, move and resize existing annotations; hold `Space` and drag to pan the canvas | | `1` | Rectangle | Box out key areas; adjustable corner radius and fill | | `2` | Ellipse | Circle / ellipse annotations | | `3` | Arrow | Directional callouts with adjustable arrow size | | `4` | Line | Straight lines | | `5` | Brush | Freehand drawing | | `6` | Highlight | Translucent marker pen to emphasize text | | `7` | Mosaic | Pixelate to blur sensitive content; adjustable block size | | `8` | Text | Add text captions with adjustable font size | Objects stay editable after drawing: pick one with the Select tool to move or resize it, or change its color, stroke width and other parameters in the properties panel. ## Image Processing The toolbar offers a group of image-processing actions to the right of the annotation tools. They operate directly on the base-image pixels, transform existing annotations in step, and are all wired into the undo history: - **Crop** (`C`) — once active, a crop box appears on the canvas; drag its 8 handles or move the whole box to pick the area you keep. The floating bar offers aspect presets (free / 1:1 / 4:3 / 3:4 / 16:9 / 9:16), or click "Auto Trim" to detect the content bounding box in one click and cut away solid-color borders. Apply to crop; press `Esc` to cancel - **Rotate** — rotate the whole image 90° clockwise / counter-clockwise, with annotations following along - **Flip** — mirror horizontally / vertically - **Resize** — enter target dimensions in a dialog with locked-aspect linking, plus 50% / 75% / 100% / 150% / 200% percentage presets; the base image is resampled at high quality - **Smart Background Transparency** — one click auto-detects the background color along the four edges (white, black or any solid color) and clears only the background connected to the borders, leaving white text and blocks inside the image untouched; edges are anti-alias feathered to remove the halo left by rounded corners and diagonals, so the exported PNG has clean edges Auto Trim and background transparency share a single **background tolerance**: drag the slider (0–128, default 16) in the "Image Processing" area of the properties panel — shown when nothing is selected — to tune sensitivity. Raise it when the background is impure or gradient, lower it to preserve detail more precisely. Transparent areas are shown on a checkerboard and exported as true transparency. ## Properties & Layers The properties panel on the right (a drawer on narrow windows) has two parts: - **Style properties** — color, custom color, fill color, opacity, stroke width; the Text tool exposes font size, rectangles expose corner radius, mosaic exposes block size and arrows expose arrow size. With an existing object selected, changes apply directly to it - **Object operations** — duplicate, delete, bring to front, send to back - **Image processing** — when nothing is selected, a background-tolerance slider (0–128) shared by Auto Trim and Smart Background Transparency The **layers list** at the bottom of the panel shows every annotation on the image (in stacking order) and supports: - Toggling visibility of individual layers (temporarily hide an annotation) - Moving layers up / down to adjust stacking order ## Viewing & Navigation - **Zoom** — the bottom bar provides zoom in / zoom out buttons, plus two presets: "Fit to Screen" (`Ctrl+0`) and "Actual Size" (`Ctrl+1`) - **Pan** — with the Select tool, hold `Space` and drag to inspect details of large images - **Undo / redo** — `Ctrl+Z` / `Ctrl+Y`, covering drawing, moving, deletion and property changes alike - The bottom bar shows the current zoom level and image dimensions in real time ## Saving & Sharing - **Copy to clipboard** (`Ctrl+Shift+C`) — copies the edited result as an image you can paste anywhere that accepts images; the button briefly shows a "Copied" confirmation - **Save** (`Ctrl+S`) — overwrites the original file path; images without a file yet (e.g. screenshots) fall back to "Save As" - **Save As** (`Ctrl+Shift+S`) — save a copy to a new path with a new name Images opened through the Launcher are edited in place: Save overwrites the original file instead of producing a copy — exactly what "quick fix" should feel like. ## Image Info & EXIF The "Info" button on the bottom bar opens the image info panel with two kinds of data: - **Basic info** — format, file size, resolution, color type, bit depth, bits per pixel, alpha channel, DPI - **EXIF data** (when the photo carries it) — manufacturer, model, lens, capture time, ISO, exposure time, aperture, focal length and 35mm-equivalent focal length, exposure program, metering mode, flash, white balance, color space, GPS coordinates and altitude, software, author, copyright and more Handy whenever you need to answer "when was this photo taken, and with what device?". ## Keyboard Shortcuts | Shortcut | Action | | --- | --- | | `V` / `1` ~ `8` | Switch between the Select tool and the 8 annotation tools | | `C` | Crop tool | | `Space` + drag | Pan the canvas | | `Ctrl+Z` / `Ctrl+Y` | Undo / redo | | `Ctrl+S` / `Ctrl+Shift+S` | Save / Save As | | `Ctrl+O` | Open image | | `Ctrl+Shift+C` | Copy to clipboard | | `Ctrl+0` / `Ctrl+1` | Fit to screen / Actual size | > **💡 macOS users** > In the table above, `Ctrl` corresponds to `Cmd` on macOS. ## Related - [Screenshot & Recording](https://yeero.ai/docs/en/capture.html) — the most common entry point for the image editor - [Launcher](https://yeero.ai/docs/en/launcher.html) — open images quickly by path or paste - [Video Editor](https://yeero.ai/docs/en/video-editor.html) — the companion tool when you need to work with video --- # Video Editor *A desktop video editing tool built into Yeero Desktop: jump straight in when a screen recording finishes, or open any local video. Multi-track timeline, 7 transitions, filters and color grading, text and subtitles, audio tools and AI-powered Smart Trim — record, edit, publish.* > **⚡ TL;DR** > The built-in video editor goes from recording to publishing: multi-track timeline, 7 transitions, filters and color grading, subtitles, audio tools and AI Smart Trim. ## Overview The video editor runs in its own window, with three areas: - **Preview** — live playback with fullscreen, fit-to-screen and A-B loop - **Timeline** — video, annotation, audio and music tracks side by side, with thumbnails and waveforms; scroll-wheel zoom and adjustable track height - **Properties panel** — select a clip or annotation to tune every parameter here A few notes on video formats: - Editing is based on **H.264-encoded MP4**; videos with other codecs are automatically transcoded to H.264 so they play and edit normally - On the first transcode or export, video processing tools (ffmpeg) are downloaded automatically — a one-time wait - Closing the window or opening a new video with unexported changes triggers a confirmation dialog, so work is never lost by accident ## Opening Videos - **After a recording** — when [screen recording](https://yeero.ai/docs/en/capture.html) finishes, choose "Open in Video Editor" and the recording loads directly - **From the menu** — click "Open video" to pick a local file, or "Open in New Window" to work on several videos in parallel - **Drag & drop** — drop a video file onto the editor window - **Launcher** — type or paste a video file path in the Launcher and press Enter ## Timeline Editing The timeline is the heart of editing. Every video clip shows thumbnails, and you can: - **Split** — press `S` to split the clip at the playhead - **Delete / ripple delete** — `Delete` leaves a gap; `Shift+Delete` ripple-deletes and later clips shift forward to fill it - **Trim** — drag the trim handles on the clip edges to cut off unwanted heads and tails - **Speed** — select a clip and adjust playback speed in the properties panel, including custom multipliers; the clip duration updates with the speed - **In / Out points** — press `I` / `O` to mark a range, `X` to clear it, `Ctrl+Delete` to delete the range - **Copy clips** — duplicate sections by copying and pasting For the view: `Ctrl+↑` / `Ctrl+↓` zoom the timeline, `Ctrl+0` fits the whole timeline, and the scroll wheel zooms too; dragging clips snaps them to adjacent clip edges. ## Transitions Add a transition between adjacent clips for smoother cuts. Seven built-in transitions: | Transition | Effect | | --- | --- | | Fade | Classic cross-fade | | Slide Left / Slide Right / Slide Up | The incoming shot slides in from that direction | | Zoom | Zoom-based transition | | Flash | White flash, great for beat-driven cuts | | Blur | Blur-through transition | Transition duration is adjustable from **0.1 to 3.0 seconds**, defaulting to 0.5s. ## Filters & Color Grading - **Filters** — Film, Vintage, Fresh, Cinema, Warm, Cool and B&W, plus "Original" to revert instantly - **Color grading** — fine-tune brightness, contrast, saturation, temperature, tint, exposure, highlights and shadows, with one-click reset ## Text & Subtitles - **Add text** — place text layers on the overlay track; adjust font, size, weight, color, opacity, stroke and shadow, with free horizontal / vertical alignment - **Text templates** — built-in layouts such as Vlog Title, applied in one click - **Text animation** — Typewriter, Fade In, Bounce and Slide Up entrances - **Subtitles** — import subtitle files or generate auto subtitles ## Video Annotations The same 9 annotation tools as the [image editor](https://yeero.ai/docs/en/image-editor.html) (Select, Rectangle, Ellipse, Arrow, Line, Brush, Highlight, Mosaic, Text) — the difference is that every annotation carries a time range: set an in point and an out point, and the annotation only appears during that span, perfect for "circle the broken button at 0:03" walkthroughs. Annotations live on a dedicated annotation track whose timing is adjustable right on the timeline. ## Audio - **Original audio** — the video's audio appears on the audio track (with waveform) and can be split, copied and deleted - **Volume control** — per-clip volume and mute / unmute, plus a global master volume - **Fade in / out** — add fade-in and fade-out to any audio clip - **Separate audio** — detach the video's audio into an independent clip - **Background music** — a dedicated music track for adding local audio files as BGM ## Smart Trim Don't want to scrub frame by frame? Smart Trim analyzes the whole video and flags redundant segments: - **Static** — segments where the picture barely changes - **Silent** — long stretches with no sound - **Duplicate** — repeated content The results list each segment with its type and duration. Choose an action per segment — **Cut**, **Speedup** or **Ignore** — or Confirm All / Ignore All. The panel shows the time you'll save in real time, and clicking any segment jumps the preview there. When you're happy, hit "Apply Cuts" and all redundant segments leave the timeline at once. ## Exporting Press `Ctrl+E` or click "Export" to open the export dialog: - **Export preset** — pick a ready-made combination - **Resolution** — keep the original or pick a target resolution - **Frame rate & format** — adjust as needed - **Output path** — choose where to save Export shows live progress, and once complete you can reveal the file in your file manager. Additionally, `Ctrl+Shift+F` captures the current frame to the clipboard anytime — handy for sharing a frame in chat. ## Keyboard Shortcuts Press `Shift+?` inside the editor to bring up the shortcuts panel anytime. | Shortcut | Action | | --- | --- | | `Space` | Play / pause | | `←` / `→` | Previous / next frame (while paused) | | `Shift+←` / `Shift+→` | Skip backward / forward 5 seconds | | `S` | Split clip at playhead | | `Delete` / `Shift+Delete` | Delete / ripple delete | | `I` / `O` / `X` | Set in point / out point / clear both | | `Ctrl+Delete` | Delete the range between in and out points | | `Ctrl+A` / `Escape` | Select all / deselect | | `Ctrl+Z` / `Ctrl+Shift+Z` | Undo / redo | | `Ctrl+↑` / `Ctrl+↓` / `Ctrl+0` | Timeline zoom in / out / fit | | `V` | Switch to the Select tool | | `Ctrl+E` | Export video | | `Ctrl+Shift+S` | Smart Trim | | `Ctrl+Shift+F` | Capture current frame to clipboard | > **💡 macOS users** > In the table above, `Ctrl` corresponds to `Cmd` on macOS. ## Related - [Screenshot & Recording](https://yeero.ai/docs/en/capture.html) — the first source of video footage - [Image Editor](https://yeero.ai/docs/en/image-editor.html) — the shared annotation tool family - [Launcher](https://yeero.ai/docs/en/launcher.html) — open video files or the editor in a keystroke --- # Disk Cleaner *Visualize disk space usage with a directory tree, auto-identify applications across 30+ categories, let AI analyze unknown folders, and safely reclaim storage — all from within Yeero Desktop.* > **⚡ TL;DR** > The Disk Cleaner scans your drives, shows every folder's size in a tree view, identifies apps automatically, and lets you clean up safely — files go to the recycle bin first, recoverable within 30 days. > **💡 Desktop only** > Disk Cleaner accesses local file systems and requires Tauri's native capabilities, so it is only available in Yeero Desktop, not on the web. ## Overview The Disk Cleaner helps you answer one question: "What is eating my disk space?" It combines a visual directory tree with built-in app recognition and optional AI analysis, so you can go from "this folder is huge" to "I know exactly what it is and whether I can delete it" in seconds. - **Multi-disk support** — scan any local, removable, network or RAM disk; switch between drives with one click - **Directory tree** — drill into folders layer by layer, each node showing its total size - **App identification** — recognized applications are tagged with their name and category (30+ built-in categories) - **AI analysis** — unknown folders can be sent to AI for classification, returning the app name, category, data type and confidence - **Safe cleanup** — files are moved to the recycle bin, never permanently deleted, giving you 30 days to recover ## Launching There are two ways to open the Disk Cleaner: 1. **Via the Launcher** — press `Alt+Space`, type "disk cleaner" or 磁盘清理, and press Enter 2. **Via the main window** — navigate to the Desktop section and select Disk Cleaner On first launch, a short onboarding dialog walks you through the three core steps: browse, identify and clean. You can dismiss it and choose not to see it again. ## Disk Overview The top bar shows a card for every detected disk: - **Drive name and letter** — e.g. "Local Disk (C:)" - **Usage bar** — a horizontal bar colored by fill level; turns red above 85% - **Used / total** — e.g. "186.3 GB / 500 GB" with a percentage - **Drive type icon** — different icons for local, USB, network, optical and RAM disks Click any card to switch the scan target to that disk. The refresh button reloads the overview. Disks with usage above 90% also trigger a warning banner below the overview bar. ## Directory Tree Once a disk is selected, the cleaner scans its top-level directories and displays them in a tree view: - **Click to expand** — each folder reveals its subdirectories with their sizes, letting you drill down to the space hogs - **Size bar** — every node shows a proportional bar indicating how much of the parent it occupies - **Scan progress** — a progress indicator at the top shows the current path, scanned file count and total size; you can cancel at any time - **Cache and refresh** — scan results are cached for fast re-entry; a timestamp shows when the data was captured, and a "Rescan" button forces a fresh scan. Background refresh keeps the tree visible while updated data arrives ## App Identification The cleaner maintains a built-in database that recognizes common applications across 30+ categories: | Category group | Examples | | --- | --- | | Browsers | Chrome, Firefox, Edge, Safari, Arc | | Development | VS Code, IntelliJ, Xcode, Android Studio, Docker | | Design | Photoshop, Figma, Sketch, Blender | | Media | Spotify, VLC, OBS, Premiere Pro | | Communication | WeChat, Slack, Discord, Teams | | System & Utility | Homebrew, npm, pip, conda | Recognized folders are tagged with an app badge showing the name and category. Each entry also carries a data-type label (cache, log, user data, config, installer, temp file, backup, media cache, package cache, etc.) and a cleanup advice level. ### AI Analysis For folders the built-in database does not recognize, you can click "AI Analyze" to let the model identify them: - **Single analysis** — analyze one folder at a time - **Batch analysis** — select multiple unknown folders and analyze them together (costs credits) - **Result details** — the AI returns an app name, category, data type and confidence level (high / medium / low) - **Feedback loop** — mark each result as correct or incorrect to improve future accuracy AI results are displayed alongside built-in identifications, so the view stays consistent. ## Cleanup Once you have identified what to remove: 1. **Select items** — check the boxes next to folders you want to clean 2. **Review the summary** — the bottom panel shows the number of selected items and the estimated space to be freed 3. **Confirm** — click "Safe Cleanup" and confirm the dialog Depending on the size and type of the cleanup, additional confirmation steps may appear: - **Large file warning** — cleaning a very large amount of data asks for explicit confirmation - **Running app warning** — if a related application is still running, you are asked to close it first - **Chat file warning** — chat-related files require typing a confirmation word, as they cannot be recovered after cleanup ## Cleanup Safety Every cleanable item is marked with one of three safety levels: | Level | Meaning | | --- | --- | | **Safe** | Cache, temp files and other data that can be removed without side effects | | **Caution** | Data that may need to be re-downloaded (e.g. package caches, media caches) | | **Not recommended** | User data or system-critical files — cleaning may cause data loss or application errors | > **⚠ Recycle Bin Protection** > All cleaned files are moved to the system recycle bin, not permanently deleted. You have 30 days to restore any file you change your mind about. After cleanup, a report dialog shows the results: how many items succeeded, how much space was freed, and any items that failed (with reasons such as path not found or system-protected path). ## Cleanup History The cleanup history panel tracks every cleanup session: - **Per-session details** — date, number of items cleaned and space freed - **Cumulative stats** — total space freed and total number of cleanups, shown in the stats bar below the disk overview - **Delete records** — remove individual history entries if desired ## Related Reading - [Launcher](https://yeero.ai/docs/en/launcher.html) — open the Disk Cleaner with a keystroke via Alt+Space - [Desktop Quick Start](https://yeero.ai/docs/en/desktop.html) — getting started with Yeero Desktop - [Files](https://yeero.ai/docs/en/files.html) — manage cloud files and automated cleanup rules --- # Files *Complete file management with upload, preview, organization, semantic search, and automated cleanup rules.* > **⚡ TL;DR** > The file library stores uploads with inline preview, folder organization, AI semantic search and automated cleanup rules. ## Files Overview The Files page provides a comprehensive file management system for all your uploaded and AI-generated content. ## Uploading Files To upload files: 1. Click **"Upload Files"** or use drag-and-drop. 2. Select files from your computer. Any format is supported, up to **512 MB** per file. 3. Choose a destination folder (optional). 4. Click **"Upload"** to start the upload. > **💡 Drag and Drop** > You can also drag and drop files directly onto the Files page to start uploading immediately. ## File Preview Click the **Preview** button on any file to open the universal file viewer: - **Images** — View with zoom in/out and reset controls. - **Videos** — Play with standard video controls. - **Audio** — Play with mute/unmute controls. - **Documents** — View supported document formats. - **Download** — Download any file to your computer. If a file type is not supported for preview, you'll be prompted to download it instead. ## Organizing Files ### View Modes - **Grid view** — Display files as thumbnails in a grid. - **List view** — Display files in a detailed list format. ### Filtering - **Source** — All, My Uploads, or AI Created. - **Type** — All Types, Images, Videos, Audios, Documents, Other. - **Sort** — Newest first or Oldest first. ### File Operations - **Rename** — Change the file name. - **Move** — Move to a different folder with a visual folder browser. - **Delete** — Permanently delete the file. - **Go to Chat** — Navigate to the conversation where the file was created. ## Semantic Search Switch from filename search to **AI semantic search** by clicking the toggle icon next to the search bar. Describe the file you're looking for in natural language, and AI will find matching files based on content similarity. ## Storage Management YeeroAI provides **1 GB of free storage**. If you exceed the free quota: - A daily credit cost applies for excess storage. - If your balance runs low, you'll see warning banners. - If overdue for more than 7 days, oldest files will be auto-deleted until storage drops to 1 GB. > **⚠ Auto-Cleanup** > If storage remains overdue for 7+ days, the system will automatically delete the oldest files until you're back within the free quota. Top up credits or manually delete files to avoid this. ## Cleanup Rules Automate file deletion with cleanup rules. Rules run daily at 2 AM. ### Creating a Rule 1. Click the **Cleanup** button on the Files page. 2. Click **"New Rule"**. 3. Configure the rule: **Rule Name** — A descriptive name (e.g., "Delete old AI-generated images"). 4. **Media Types** — Filter by image, video, audio, etc. 5. **File Sources** — Filter by uploaded or AI-created. 6. **Uploaded more than** — Minimum days since upload (7-3650 days). 7. **File size ≥** — Minimum file size in MB. 8. **Path Scope** — Use wildcards to target specific folder paths (e.g., `/photos/**`). ### Managing Rules - **Enable/Disable** — Toggle rules on and off. - **Edit** — Modify rule conditions. - **Delete** — Remove a rule permanently. - **Preview** — Re-run the preview to see current matching files. - **Delete Now** — Manually execute the rule immediately. ## Batch Operations The file library supports drag-to-select rectangle selection for quick multi-file selection and batch actions. Press and hold the left mouse button on an empty area of the list to draw a selection rectangle. Release to highlight all files in range, then use the floating action bar to **batch download**, **batch delete**, or **batch move to folder**. Press `Esc` to exit multi-select mode. Batch download supports both files and folders. Files inside folders are downloaded flat in the browser, while the original directory structure is preserved in the Tauri desktop app. --- # Folders *Nested folder hierarchy with pinning, sorting, move operations, and parameter inheritance for chats, apps, and messages.* > **⚡ TL;DR** > Folders organize chats, apps, prompts and knowledge in nested hierarchies, with pinning, sorting, batch moves and folder-level parameter inheritance. ## Folders Overview Folders in YeeroAI help you organize chats, apps, messages, memories, and prompts in a nested hierarchy. The folder system supports: - **Three folder types** — Chat Folders, App Folders, and Message Folders. - **Nested structure** — Create subfolders within folders for deep organization. - **Parameter inheritance** — Set models, tools, memory, and generation parameters at the folder level, and child items inherit them. ## Navigating Folders ### Breadcrumb Navigation The breadcrumb at the top shows your current folder path. Click any segment to navigate up the hierarchy. Use the folder search in the breadcrumb to find subfolders quickly. ### Sidebar Navigation The Folders page in the sidebar provides access to all three folder types with search, sort, and filtering options. ## Folder Operations | Operation | Description | | --- | --- | | New Folder | Create a new folder with name and optional description. | | Edit | Modify folder name and description. | | Pin / Unpin | Pin folders to the top of the list for quick access. | | Move | Move a folder into another folder. | | Delete | Permanently delete a folder and all its contents. | | Enable/Disable Memory | Toggle AI memory for all items in the folder. | | Enable/Disable Prompt | Toggle prompt templates for the folder. | ## Parameter Inheritance Folders support parameter inheritance, allowing you to set defaults that all child items (chats, subfolders) inherit: - **Models** — Default model selections for conversations in this folder. - **Tools** — Default tool selections. - **Memory** — Enable or disable AI memory for the folder. - **Context Summary** — Use summaries instead of full content. - **Generation Parameters** — Temperature, sampling, context window, etc. ### Inheritance Modes - **Inherited** — Uses the parent folder's settings (default). Marked with an "Inherited" badge. - **Custom** — Uses custom settings specific to this folder. Click "Switch to Custom" to override. - **Reset to Inherit** — Return to inheriting from the parent folder. ## Content in Folders Folders can contain the following types of content: | Type | Description | | --- | --- | | Folder | Nested subfolder. | | Chat | A conversation with branches and messages. | | Message | An individual message (user or AI). | | Memory | An AI-generated or manually created memory entry. | | Prompt | A reusable prompt template. | | App | An AI-powered application. | ## Folder Limits | Limit | Value | | --- | --- | | Maximum folders | 1,000 | | Folder name uniqueness | Names must be unique within the same parent folder. | | Cannot move into descendant | A folder cannot be moved into one of its own subfolders. | ## Batch Operations Reference lists inside folders support drag-to-select rectangle selection, letting you quickly select multiple items and perform batch actions. Press and hold the left mouse button on an empty area to draw a selection rectangle. Release to highlight all items in range, then use the floating action bar to **batch delete** or **batch move**. Press `Esc` to exit multi-select mode. --- # Prompt Templates *Create and manage reusable prompt templates with enable/disable toggle. Insert any prompt into conversations with one click.* > **⚡ TL;DR** > Prompt templates save your best prompts for one-click reuse in any chat, with enable/disable toggles and folder organization. ## What Are Prompt Templates? Prompt Templates are pre-written prompts that you can quickly insert into any conversation. They help you: - **Save time** — Reuse common prompts without retyping. - **Ensure consistency** — Use standardized prompts across conversations. - **Share best practices** — Create prompts for common tasks and workflows. ## Creating a Prompt 1. Navigate to **Settings → Prompts**. 2. Click **"New Prompt"**. 3. Fill in the form: **Prompt Name** — A descriptive name for the prompt. 4. **Prompt Content** — The prompt text. Markdown is supported. 5. **Folder** — Choose which folder to save the prompt in. ## Managing Prompts ### Enable / Disable Toggle prompts on and off. Disabled prompts won't appear in the quick-insert menu but are not deleted. ### Search and Filter - **Search** — Search prompts by name. - **Status filter** — Show All, Enabled, or Disabled. - **Sort** — Recently Updated or Earliest Created. ### Edit and Move - **Edit** — Modify the prompt name, content, or folder. - **Move** — Move the prompt to a different folder. - **Delete** — Permanently delete the prompt. ### Preview Click **"Preview"** to see how the prompt content renders with Markdown formatting. ## Batch Operations ### Batch Move 1. Click **"Select"** to enter selection mode. 2. Choose the prompts you want to move. 3. Click **"Batch Move"** and select the target folder. ### Batch Delete You can batch delete prompts in two ways: - **Delete All Disabled** — Remove all disabled prompts at once. - **Select and Delete** — Select specific prompts and batch delete them. ## Using Prompts in Conversations When composing a message in a chat, you can insert enabled prompts from your library. The prompt content will be added to your message input, ready to send. ## Preset Role Prompt Library Beyond creating your own prompts, YeeroAI ships with a built-in **official preset role prompt library** covering 242 carefully written roles across 25 categories (bilingual), so you can get professional-grade prompts without writing from scratch. ### Browse and Import 1. Go to **Settings → Prompts** and switch to the **"Preset Library"** tab. 2. Filter by category or search by keyword (matches names, descriptions, and aliases). 3. Click **"Preview"** to read the full role prompt. 4. Click **"Use This Preset"**, choose a folder, and it is imported into **"My Prompts"** in one click. ### Official Updates and Sync Imported presets record their source and official version. When an updated official version is published, the prompt card shows an **"Official update available"** marker; click **"Sync to Latest Official"** to update its content to the latest official version. > **⚠️ Syncing overwrites local edits** > Syncing **overwrites** your current name and content with the latest official version and cannot be undone. Back up first if you have made your own edits. Once imported, a preset is just a normal **My Prompt**: you can freely edit, move, enable/disable, and delete it, and editing never affects the official original. ## AI-Assisted Authoring When creating or editing a prompt, you can use **AI Assist** to generate or refine content quickly instead of starting from a blank page. - **AI Generate** — Describe the role or purpose you want, and AI generates a structured prompt for you. - **Rewrite / Polish / Expand / Simplify** — Rewrite, polish, expand, or simplify existing content. - **Model selection** — Choose the model used for assistance, or let it be selected automatically. AI Assist results appear in a preview first; once you are satisfied, click **"Apply"** to write them into the prompt content, and you can keep editing manually. > **ℹ️ About usage** > AI Assist consumes credits per use; see the in-app prompt for details. ## Prompt Limits | Limit | Value | | --- | --- | | Maximum prompts | 1,000 | | Name must be unique | Prompt names cannot be duplicated. | --- # Settings *Manage theme, language, default models, memory, background streaming, cloud sync, and all application preferences.* > **⚡ TL;DR** > Settings centralize everything about how YeeroAI behaves: theme, language, default models, memory, context size, background streaming and cloud sync. ## Settings Overview Access Settings from the sidebar. Settings are organized into sections: - **System** — Theme, language, end-to-end encryption key, and basic configuration. - **General** — Auto chat naming, context window, memory, default models, reasoning effort. - **Background Streaming** — Continue AI generation when navigating away. - **Cloud Sync** — Sync settings and app data across devices. - **Prompts** — Manage prompt templates. - **Memories** — Manage AI memory entries. ## System Settings ### Theme Switch between **Light** and **Dark** themes. Your preference is saved and persists across sessions. ### Language Switch the interface language between **English** and **中文**. The interface updates immediately. ### End-to-End Encryption Key Configure a universal end-to-end encryption key used by all encrypted sync scenarios (cross-device clipboard sync, Agent messages, etc.). The key is stored only in your device's OS keychain; leaving it empty disables encryption. - **One-click generate** — let the app generate a strong key for you instead of remembering one yourself. - **Keep it identical across devices** — to decrypt synced encrypted content (e.g. clipboard entries) on another device, configure the same key there. ## General Settings | Setting | Description | | --- | --- | | Auto-generate Chat Name | Automatically generate chat titles from conversation content. | | Context Window Size | Number of historical messages retained in conversations. Must be a number within the valid range. | | Enable Memory | Allow AI to remember key information from conversations. Can be overridden per chat or folder. | | Default Models | Set default models for all new chats. Select up to 10 models. | | Reasoning Effort | Default reasoning effort level: None, Minimal, Low, Medium, High, or Extreme High. | | Context Summary | Use message summaries instead of full content when building context. Reduces token usage but may lose some detail. | ## Background Streaming When **Background Streaming** is enabled: - AI generation continues even if you navigate away from the chat. - Responses are preserved after page refresh or network interruption. - When you return to the chat, you'll receive any remaining content. > **💡 Resilient** > Background streaming handles network interruptions gracefully. You won't lose any generated content even if your connection drops temporarily. ## Cloud Sync ### Sync Project Settings Sync your theme and other local preferences to the cloud, so they're consistent across all your devices. ### Sync App Data Sync AI App runtime storage data to the cloud. You can see which apps have sync disabled and re-enable them individually. ## Memories Management The **Memories** section in Settings lets you manage all AI memory entries: - **View all memories** — Including auto-generated and manually created ones. - **Search** — Search memories by name or content. - **Filter by status** — Enabled, Disabled, Embedding, or Forgotten. - **Filter by trait** — Fixed, Very Stable, User Habit, User Preference, Mutable, or Weak. - **Enable / Disable** — Toggle individual memories. - **Edit** — Modify memory name and content. - **Move** — Move to a different folder. Note: auto-generated memories from chats cannot be moved. - **Delete** — Permanently delete a memory. - **Batch operations** — Batch move and batch delete memories. ## Batch Operations The Prompt Templates list and Memories list in Settings both support drag-to-select rectangle selection for quick multi-item selection and batch actions. Press and hold the left mouse button on an empty area to draw a selection rectangle. Release to highlight all items in range, then use the floating action bar. Press `Esc` to exit multi-select mode. | List | Available Batch Actions | | --- | --- | | Prompt templates list | Batch delete, batch move to folder | | Memories list | Batch delete, batch move to folder | ## Account Settings From the Account page, you can manage: - **Profile** — Edit nickname, username, bio, and upload an avatar. - **Account Security** — Bind/unbind email, Google, and GitHub accounts. - **Sign Out** — Log out and clear local login state. ### Login Methods You can bind multiple login methods for security: - **Email** — Bind or change your email address with verification. - **Google** — Link or unlink your Google account. - **GitHub** — Link or unlink your GitHub account. > **⚠ Keep At Least One Method** > You must keep at least one login method bound to your account. The system will prevent you from unlinking the last method. --- # Billing & Credits *Simple, transparent pay-as-you-go billing — credits on signup, per-reply cost visible in real time, top up anytime, cancel anytime.* > **⚡ TL;DR** > YeeroAI runs entirely on pay-as-you-go credits: 1,000 free at signup (valid 30 days), per-reply cost shown in real time, one-time credits never expire, and purchases are refundable pro-rata within 24 hours. ## How Credits Work Every AI capability on YeeroAI is settled in a single currency — **credits**. There is no mandatory monthly fee: you pay only for what you use. - **1,000 credits on signup** — valid for 30 days, no credit card required. - **Cost visible in real time** — every AI reply shows exactly how many credits it consumed. Nothing is charged behind the scenes. - **Free models cost nothing** — turn on "Free only" in the model picker to use models that consume zero credits. - **Balance always at hand** — check your balance, transaction history and bills under **Account → Credits**. - **When you run out** — AI requests fail with a top-up prompt. Your existing chats, files and knowledge entries are never deleted because of it. > **💡 What determines the cost** > Each request is priced by the model used and the length of its output. When one message is sent to several models in parallel, each model is billed separately. ## Credit Validity Validity depends on where the credits came from. Expired credits can be neither restored nor refunded: | Source | Validity | Notes | | --- | --- | --- | | Signup bonus | 30 days | The 1,000 welcome credits; any unused portion is deducted as an "Expired" entry. | | One-time purchase | Never expires | Credit packs stay valid indefinitely, so you can stock up as needed. | | Subscription monthly credits | 30 days per cycle | Issued with each billing cycle and forfeited once the subscription ends or is canceled. | | Invitation rewards | Never expire | Credits earned by inviting friends stay valid indefinitely. | | Promotional rewards | Per campaign terms | Validity is stated by the campaign and shown on each credit record. | The exact expiry of every grant is listed in the "Expires" column under **Account → Credits → Earned**; permanent credits are labelled **"Never expires"**. ## Earning Credits | Method | Description | | --- | --- | | **Registration Bonus** | 1,000 credits on sign-up (valid for 30 days). | | **Purchase** | Buy credit packs or a subscription plan — see Pricing Plans. | | **Invitation Reward** | Earn 10% of your friends' registration credits and 10% of every credit purchase they make — uncapped and never expiring. See Invitation Rewards. | | **Activity Reward** | Credits from special promotions and official campaigns. | ### Reading your earn history Under **Account → Credits → Earned**, every credit grant is fully traceable: | Column | Meaning | | --- | --- | | Earn Method | Registration Bonus, Purchase, Invitation Reward or Activity Reward. | | Time | When the credits landed in your account. | | +credits | How many credits this entry added. | | Balance Before | Your balance before this grant, for line-by-line reconciliation. | | Remaining | How much of this specific grant is still unused. | | Expires | When this grant expires, or "Never expires". | | Remark | Extra context such as the campaign name or order info. | | Action | Purchased credits can be refunded from here. | Filter by date range (up to one year back); the list loads the next page automatically as you scroll. ## Spending Credits ### AI calls | Method | Description | | --- | --- | | **Chat Message** | Every AI reply, priced by model and output length — the cost is shown right on the reply. | | **App Generation** | Generating code and content for AI apps. | | **App Discussion** | Discussion threads inside an AI app. | | **Agent Model Call** | Model calls made automatically in Agent mode. | | **Cloud Agent Container Runtime** | Container resources consumed by cloud Agent runs. | | **Disk Cleaner Analysis** | AI analysis in the desktop Disk Cleaner — the exact credit cost is shown before it runs. | ### Memory & knowledge | Method | Description | | --- | --- | | **Generating Memory** | Distilling knowledge entries from your conversations. | | **Matching Memory** | Matching relevant memory context for a new message. | ### Storage | Method | Description | | --- | --- | | **File Storage Fee** | Once your files exceed the **1 GB** free quota, the excess is billed at **50 credits / GB / day** — see Storage Billing. | | **Cloud Document Storage Fee** | Once cloud documents created by AI apps exceed the **10 MB** free quota, the excess is billed at **5 credits / 10 MB / day** — see Storage Billing. | ### Other | Method | Description | | --- | --- | | **Expired** | Credits deducted once past their validity period. | | **Refund** | Remaining credits reclaimed after a successful refund. | ### Reading your spend history Under **Account → Credits → Spent**, every deduction shows the spend method, time, credits deducted, balance before, balance after and a remark — handy for auditing what a single chat or app really cost. Date-range filtering and infinite scroll work exactly as above. ## Storage Billing YeeroAI keeps **two independent storage quotas**: your file library and AI app cloud documents. They do not share capacity and are billed separately — anything above either quota is deducted from your credit balance daily. ### Quotas and rates | Storage type | Free quota | Overage rate | Spend method | Where to check usage | | --- | --- | --- | --- | --- | | **File library** | 1 GB | 50 credits / GB / day | File Storage Fee | The storage status bar at the top of the Files page | | **AI app cloud documents** | 10 MB | 5 credits / 10 MB / day | Cloud Document Storage Fee | The storage status bar at the top of the AI Apps page | The table above bills your **account-level storage totals**. Keep it distinct from one other limit: each AI app also has a "1 MB per app" cap on cloud-synced runtime data. Going over that cap only pauses cloud sync for that app until you clear its data — it **never incurs a credit charge**. See [Cloud Sync](https://yeero.ai/docs/en/ai-apps.html#cloud-sync). ### How the daily charge is calculated Only the portion **above the free quota** is billed. It is scaled proportionally and **rounded up**, with a floor of 1 credit: **Daily credits = Excess capacity ÷ Free quota × Rate** | Actual usage | Excess | Calculation | Charged per day | | --- | --- | --- | --- | | File library 800 MB | None | — | 0 credits | | File library 1 GB + 1 MB | 1 MB | 1 MB ÷ 1 GB × 50 ≈ 0.05, rounded up | 1 credit | | File library 1.5 GB | 0.5 GB | 0.5 × 50 | 25 credits | | File library 3 GB | 2 GB | 2 × 50 | 100 credits | | Cloud documents 8 MB | None | — | 0 credits | | Cloud documents 15 MB | 5 MB | 5 MB ÷ 10 MB × 5 = 2.5, rounded up | 3 credits | | Cloud documents 40 MB | 30 MB | 3 × 5 | 15 credits | ### Billing rules - **Settled daily** — while you are over quota, the charge is deducted from your credit balance once per day; each of the two storage types produces its own spend record, so you can audit them separately. - **Nothing charged within quota** — once you delete files or documents and usage drops back inside the free quota, no further storage fees accrue. - **Always visible** — the status bar shows the current daily charge directly, as "25 credits/day" for the file library and "Current +3 credits/day" for cloud documents, so you never have to do the maths. - **Days of balance left** — the status bar estimates how many days your balance still covers (current balance ÷ daily charge) and turns into a red "Balance low (< 3 days)" warning when fewer than three remain. > **💡 The cloud document quota is smaller than you might expect** > AI app cloud documents come with just **10 MB** free, entirely separate from the file library's 1 GB — the two cannot borrow from each other. Apps that run for a long time and keep producing documents go over quota easily, so keep an eye on the storage status bar at the top of the AI Apps page; its tooltip shows both the rate and your current daily charge. ### What each on-screen state means | State | What you see | What it means | | --- | --- | --- | | Within quota | "Storage OK", "X / 1 GB" | No storage fee at all. | | Over quota | "Over free quota", "X over", "X credits/day" | Daily billing has started; the current daily cost is shown live. | | Balance running low | "Balance low (< 3 days)", "Balance lasts X days" | How many days your balance still covers at the current daily cost — a prompt to top up. | | Insufficient balance | "Storage overdue" | Your balance can no longer cover the storage fee, so the account is in arrears. | ### What happens when you are in arrears Arrears and the 7-day grace period are tracked at **account level**, so both storage types share the same timeline: 1. **Arrears begin** — a "Storage overdue" banner appears at the top of the Files page, showing how many days you have been overdue. 2. **Grace warning** — the banner tells you that unless credits are topped up within the remaining days, the system will start auto-deleting from your oldest uploads until storage drops to 1 GB. 3. **More than 7 days overdue** — automatic cleanup begins: the file library is trimmed from your oldest uploads until it is back inside the 1 GB free quota, and cloud documents are trimmed until they are back inside the 10 MB free quota. > **⚠ Automatic cleanup cannot be undone** > Files and documents removed by automatic cleanup cannot be recovered. When you see the arrears banner, either **top up credits** or **manage your content** right away — both actions are offered directly on the banner. > **⚠ About this quota and rate snapshot** > The quotas and rates above are a snapshot taken on **2026-09-04** and may change as our operations evolve. For what is actually charged, trust the daily figure shown live on the status bar and the itemised entries under **Account → Credits → Spent**. ## Pricing Plans Plans come in two shapes: **subscriptions** (monthly credits at a lower unit price) and **one-time purchases** (credits you keep forever). They work well together — a subscription covers your baseline usage, one-time packs cover the spikes. ### Subscription tiers | Tier | Per month | Credits / month | Per credit | | --- | --- | --- | --- | | **Pro** | $5.00 | 37,500 | ≈$0.000133 | | **Pro Plus** | $10.00 | 75,000 | ≈$0.000133 | | **Ultra** | $20.00 | 150,000 | ≈$0.000133 | ### One-time purchases | Payment channel | Unit price | Credits per unit | Per credit | | --- | --- | --- | --- | | Paddle (international cards, billed in USD) | $1.00 | 6,000 | ≈$0.000167 | | WeChat Pay (billed in CNY) | ￥1.30 | 1,200 | ≈￥0.00108 | | Alipay (billed in CNY) | ￥1.30 | 1,200 | ≈￥0.00108 | > **💡 Subscriptions are roughly 20% cheaper per credit** > Subscription tiers work out at about $0.000133 per credit versus $0.000167 for one-time packs. Choose a subscription if your usage is steady; lean on one-time packs if it fluctuates, so credits never sit idle until they expire. > **⚠ About this price snapshot** > The tables above are a snapshot taken on **2026-09-04**. Tiers, prices and credit amounts are served live and may change as our operations evolve — always trust the [pricing section on the homepage](https://yeero.ai/en#pricing) and **Account → Credits → Buy Credits**. ### Get started free $0 — 1,000 credits on registration (valid for 30 days), no credit card required. ### Every tier includes - Multi-model parallel chat - Knowledge base & files - Fast AI app creation ## Buying Credits ### Where to buy - **Pricing section on the homepage** — if you are signed out, clicking a tier first prompts you to sign up or log in. - **Account → Credits → Buy Credits** — top up anytime while signed in; the arrears banner on the Files page also offers a "Top Up Credits" shortcut. ### Purchase steps 1. Pick a tier in the pricing carousel — browse tiers left and right; the focused one is badged "Recommended". 2. For one-time packs, adjust the **Quantity**; the panel live-updates "X credits · total price" and the per-credit price. 3. Choose a **Payment Method**: Paddle, or WeChat / Alipay — WeChat Pay and Alipay can be switched within the same tier. 4. Complete the payment and your credits arrive. ### Card payment (Paddle) Clicking subscribe or buy opens the Paddle checkout; follow the prompts to pay. On success the checkout closes automatically and your balance and subscription status refresh. Best for international card holders, billed in USD. ### QR code payment (WeChat Pay / Alipay) Choosing WeChat Pay or Alipay opens a QR code dialog showing the credits you will receive and the amount due: - Scan the code with the matching app to pay. - **The QR code is valid for 5 minutes**, and the page polls the payment status for you — no manual refresh needed. - On success you will see "Payment successful! Credits added." - If you see "Payment failed or timed out", or the code has expired, simply start the purchase again. > **💡 Paid but no credits yet?** > Payment callbacks are occasionally delayed — refresh the page or reopen the Credits tab first. If the order still reads "Pending" in your **Billing** history, the payment has not been confirmed yet, so please do not pay twice. If it stays stuck, reach us via **Account → Feedback** and include the order time. ## Managing Your Subscription A subscription runs in 30-day billing cycles, and that cycle's credits are issued when it starts. ### Subscription statuses | Status | Meaning | | --- | --- | | **Active** | Running normally — it auto-renews each cycle and issues credits. | | **Trialing** | In a trial period; nothing is charged before it ends. | | **Paused** | Temporarily stopped; per-cycle billing resumes once reactivated. | | **Past Due** | A renewal charge failed. Update your payment method promptly, or the subscription may not continue. | | **Canceled** | Stops at the end of the current billing period; everything keeps working until then. | ### The current subscription card When you are subscribed, a card at the top of **Account → Credits** shows the plan name, status, credits per month, current billing period, next billing date and expiry date. ### Upgrading 1. Select a higher tier in the Buy Credits panel. 2. The system checks whether your current subscription allows it: if so the button reads "Upgrade", otherwise the reason is shown directly. 3. After payment you will see "Upgrade successful. Your new billing cycle starts from the payment date." ### Canceling 1. Go to **Account → Credits** and click **"Cancel Subscription"** on the current subscription card. 2. The confirmation dialog states three things: you keep access for the rest of the current period, you will not be charged for the next one, and the exact expiry date. 3. Click **"Confirm Cancel"** — or "Keep Subscription" to stay. > **💡 Canceling does not cut you off** > After canceling, the status becomes "Canceled", but everything keeps working until the current period ends, including the credits already issued this cycle. There is no further auto-renewal, and that cycle's credits lapse per the validity rules. ## Credit & Billing Records Everything about your credits and bills lives under **Account → Credits** (URL: `/account?tab=credits`): | Area | What it holds | | --- | --- | | Credit Balance | Your total available credits, with a manual refresh. | | Current Subscription | Shown when subscribed — status, next billing and expiry. | | Buy Credits | Opens the pricing panel to top up or upgrade anytime. | | Earned | Every credit grant, including the refund entry point. | | Spent | Every credit deduction. | | Billing | Payment records for all orders and subscriptions. | ### Billing columns | Column | Meaning | | --- | --- | | Type | "One-Time" (a single credit purchase) or "Subscription" (a recurring payment). | | Channel | Paddle, WeChat Pay or Alipay. | | Credits | Credits granted by that record. | | Amount | What you actually paid, with its currency. | | Status | Order status — Pending / Completed / Refunded / Failed; subscription rows show the subscription status instead. | | Billing Period | Cycle start and end dates, for subscription records. | | Time | When the order was placed or charged. | All three sub-tabs support date-range filtering (up to one year back) and load more as you scroll. ## Refunds Credit purchases are refundable, but there is a strict time window and a specific calculation. Please read this before you submit. > **⚠ Four hard refund rules** ### How the refund amount is calculated **Refund = Amount actually paid × (Remaining credits ÷ Credits purchased)** Example: you paid $10 for 1,000 credits and used 300, leaving 700. The refund is $10 × (700 ÷ 1,000) = **$7.00**; the $3.00 behind the 300 used credits is not refunded. - Expired credits do not count towards the remaining balance. - If a coupon or discount was applied, the refund is based on the **amount actually paid**, not the list price. - Amounts are rounded to the smallest currency unit (cents for USD). ### Eligibility All of the following must hold: 1. The request is submitted within 24 hours of the purchase. 2. Some of the purchased credits are still unused. 3. The credits came from a direct payment — not a promotion, giveaway or invitation reward. 4. The account is in good standing, not restricted or banned for breaching the Terms of Service. 5. The account has never requested a refund before. Conversely, refunds are refused when the purchase is older than 24 hours, the credits are fully used, the credits were free, the account is banned, or a refund has already been processed. ### How to request one 1. Go to **Account → Credits → Earned**. 2. Find the "Purchase" record and click **"Apply Refund"**. The button stays disabled when the record is not eligible. 3. Check the details: Order Information, Purchased Credits, Used, Remaining Credits, Estimated Refund, and the "Original Payment Method" note. 4. Click **"Confirm Refund"**. On success you will see "Refund Request Submitted — your request is being processed". The record then reads **"Refund Processing"** and finally **"Refunded"**; the reclaimed credits appear in your spend history as "Refund". ### Timing and statuses | Stage | Duration | | --- | --- | | Internal review | 1-3 business days | | Settlement by the payment provider | 5-15 business days, varying by provider and region | A refund is **Processing** (under review), **Completed** (approved and handed to the payment provider) or **Failed**. Refunds always go back to the **original payment method** — we cannot switch channels or convert them into account balance. The full terms live in the [Refund Policy](https://yeero.ai/policy/refund/). ## Invitation Rewards Found under **Account → Invitations**: - **Your Invitation Code** — click "Copy Code" and share it with friends. - **Invite Link** — share your personal link; the code is applied automatically when friends sign up through it. - **10% reward** — you earn 10% of a friend's registration credits, plus 10% of the credits they buy, every single time. - **No limit** — invite more, earn more, and invitation credits never expire. - **Your stats** — the page shows Invitees, Rewards Earned, and the Invited Users list (with scroll-to-load-more). ### Invitation rules 1. Each user has a unique invitation code to share with friends. 2. When a friend registers with your code, you earn 10% of their registration credits. 3. When a friend purchases credits, you earn 10% of their purchased credits. 4. No limit on invitation rewards — invite more, earn more. 5. Credits earned from invitations never expire. 6. Fraudulent invitations are strictly prohibited — violators will have their accounts banned. Invitation credits are labelled "Invitation Reward" in your earn history, and are not refundable. ## Still Have Questions - [FAQ](https://yeero.ai/docs/en/faq.html) — high-frequency answers on credits, models, privacy and troubleshooting. - **Account → Feedback** — file a bug report or feature request with screenshots, and track its status on the same page. - [Refund Policy](https://yeero.ai/policy/refund/) and [Terms of Service](https://yeero.ai/policy/terms/) — the full legal text. --- # FAQ *Answers to frequently asked questions about using YeeroAI.* > **⚡ TL;DR** > Straight answers to the most common YeeroAI questions — accounts, credits, models, core features, desktop and CLI, privacy, storage, and troubleshooting, all in one place. ## Account & Login ### How do I sign up? Visit yeero.ai and sign up with email (verification-code login), Google, or GitHub. You'll receive **1,000** free credits instantly (valid for 30 days). ### Can I use a test account? Yes, click "Try test account" on the login page. Test accounts have limited functionality — some operations (such as payments) are restricted, so you can explore before committing. ### How do I change my email? Go to Account → Account Security, then click "Bind" or "Switch" next to the email. You'll need to verify your identity with your current login method first. ### Can I link multiple login methods? Yes, you can link email, Google, and GitHub to the same account. We recommend linking multiple methods for better security, but you must keep at least one method linked. ### How do I sign out? Go to Account → Sign Out, then click "Sign Out" for the current device or "Sign Out All Devices" to force-logout every signed-in device. ### What if I don't receive the email verification code? Check your spam folder first — the code email may land there. If it still hasn't arrived, wait for the countdown to finish and click "Resend". If the problem persists, contact us via Account → Feedback. ### How do I delete my account? For data-security reasons, account deletion is handled manually. Submit a deletion request via Account → Feedback (mark it "Account Deletion"), and we will delete or anonymize all your data within 30 days. See our [Privacy Policy](https://yeero.ai/policy/privacy/) for details. ## Credits & Billing ### How many free credits do I get? You get **1,000** credits on registration (valid for 30 days). Inviting friends earns additional credits — invitation credits never expire. ### How are credits consumed? Credits are used for AI conversations, memory generation and matching, AI app generation and discussion, Agent and cloud runs, and storage overage. The amount depends on the model and output length, and the exact cost is shown on each AI response. See [Spending Credits](https://yeero.ai/docs/en/billing.html#spending) for the full list. ### Do credits expire? One-time purchased credits never expire. Subscription credits may expire when your subscription ends. The 1,000 registration credits are valid for 30 days — check the "Expires" column under Account → Credits → Earned for other gifted credits. See [Credit Validity](https://yeero.ai/docs/en/billing.html#expiry). ### Can I get a refund? Yes. Go to Account → Credits → Earned and click "Apply Refund" on a purchase record. Refunds must be requested within **24 hours** of purchase, are calculated proportionally to remaining credits, are limited to one per account, and cannot be withdrawn once submitted. Internal review takes 1-3 business days, and settlement back to your original payment method takes a further 5-15 business days. See our [Refund Policy](https://yeero.ai/policy/refund/) and [Refunds](https://yeero.ai/docs/en/billing.html#refunds). ### What payment methods are supported? Paddle (international credit cards) and WeChat Pay / Alipay (for users in China). All payments are processed by licensed third parties; YeeroAI never stores full card numbers. ### What's the difference between a subscription and one-time purchase? A subscription provides a fixed monthly credit allowance, ideal for ongoing use; one-time purchases are pay-as-you-go and the credits never expire. You can use both together, and subscriptions can be canceled anytime — you keep access until the end of the current billing cycle. ### Where can I see my credit usage details? Go to Account → Credits → Spent to see every deduction (conversations, memories, apps, storage). The Billing sub-tab on the same page lists all your orders and subscription history. ### How do invitation rewards work? You earn **10%** of your friend's registration credits and **10%** of every credit purchase they make. Invitation rewards are uncapped and never expire. Get your unique code and link in Account → Invitations. See [Invitation Rewards](https://yeero.ai/docs/en/billing.html#invitations). ## AI Models ### How many models are available? **300+** models from 10+ providers including OpenAI, Anthropic, Google, Meta, and Mistral, all accessible through OpenRouter. ### Are there free models? Yes, toggle "Free Only" in the model selector to see free models. Each call to a free model costs **1** credit. ### Can I use multiple models at once? Yes, select up to **10** models in the same conversation for parallel comparison — answers appear side by side for easy evaluation. ### What modalities are supported? Text, image, audio, video, and file input/output, depending on the capabilities of the selected model. ### How do I choose the right model? For everyday chat, pick a general model (such as GPT or Claude); for code, choose a model strong at programming; for long documents, check the context length. Each model's detail page lists capability tags, context window, and credit price so you can compare. ### What if a model is unavailable? Models can be temporarily unavailable due to upstream provider maintenance. Switch to another model to continue; if the issue persists, let us know via Account → Feedback. ## Core Features ### What is branching thinking? Branching gives every conversation Git-style branches: fork a new path from any message to explore different directions, merge to keep the best ideas, and roll back to any node — the main thread is never affected. See [Branching Thinking](https://yeero.ai/docs/en/branching.html). ### What is Follow-Up? Select any passage in an AI reply and launch a follow-up in one click. Follow-ups unfold in a separate thread, support multi-level drilling, and leave the main conversation's context intact. See [Follow-Up](https://yeero.ai/docs/en/followup.html). ### What are AI Apps? Describe an idea in natural language and AI generates a runnable HTML or Python app. Apps support patch-based incremental iteration, full version history, live preview, and built-in discussion. See [AI Apps](https://yeero.ai/docs/en/ai-apps.html). ### How does the Knowledge Base work? AI automatically extracts key knowledge from your conversations, builds a vector index, and injects relevant context into future chats. You can browse, search, and manage entries by tag. See [Knowledge Base](https://yeero.ai/docs/en/knowledge-base.html). ### What are AI Memories? Memories are persistent knowledge entries AI distills from your conversations. They are automatically matched and injected as context in later chats, so the AI understands you better over time. View, edit, and delete them in Settings → Memories. ## Desktop & CLI ### Is there a desktop app? Yes. Yeero Desktop is built with Tauri and supports macOS (Apple Silicon / Intel), Windows (x64 / ARM64), and Linux (AppImage / deb), with native windows, OS notifications, and in-app auto-updates. See [Desktop](https://yeero.ai/docs/en/desktop.html). ### What is the Launcher? The Launcher is the desktop app's global summoner, triggered by `Alt`+`Space`. It ships with 16 processors covering calculation, app launching, search, and AI chat. See [Launcher](https://yeero.ai/docs/en/launcher.html). ### Does the desktop app support screenshots and screen recording? Yes. The desktop app offers smart window detection, scrolling-screenshot stitching, and microphone/webcam picture-in-picture recording. Screenshots can go straight into the image editor for annotation. See [Screenshot & Recording](https://yeero.ai/docs/en/capture.html). ### Is there a CLI? Yes. Install the yeero CLI with one command: on macOS / Linux run `curl -fsSL https://yeero.ai/cli/install.sh | sh`; on Windows run `irm https://yeero.ai/cli/install.ps1 | iex`. See [CLI Quick Start](https://yeero.ai/docs/en/cli-getting-started.html). ### Does the CLI share data with the web app? Yes. After signing in, the CLI shares the same account with the web and desktop apps, and conversation history syncs in real time. ## Privacy & Security ### Is my data encrypted? Yes. All data is encrypted in transit with **TLS 1.2+** and at rest with **AES-256**. Your private data is never sold or shared with any third party. ### Can I delete my data? Yes, you can delete individual conversations, messages, files, and knowledge entries at any time. After account deletion, all data is removed or anonymized within **30 days**. Deletion is permanent and cannot be undone. ### What about AI training? YeeroAI does not use your data to train AI models. However, your queries are sent to underlying model providers (such as OpenAI, Anthropic, and OpenRouter) to generate responses, and that data is subject to those providers' privacy policies. See our [Privacy Policy](https://yeero.ai/policy/privacy/). ### What is the end-to-end encryption key? You can configure a universal E2E key in Settings → End-to-End Encryption Key for encrypted sync scenarios (such as cross-device clipboard sync). The key is stored only in your local system keychain; leaving it blank disables encryption. To decrypt synced content across devices, configure the same key on each one. ### Where is my data stored? YeeroAI stores your data on cloud infrastructure certified under SOC 2 Type II and ISO 27001, encrypted in transit and at rest, and protected by strict access controls. See our [Privacy Policy](https://yeero.ai/policy/privacy/) for details. ## Storage & Files ### How much storage do I get? **1 GB** of free storage. Beyond the free quota, extra storage incurs a daily credit fee. ### What happens if I exceed my storage? You'll see warning banners. If storage stays overdue for more than **7 days** and your credit balance is insufficient, the oldest files are auto-deleted until you're back within the 1 GB free quota. ### What file types can I upload? Any format is supported, up to **512 MB** per file. Images, videos, audio, and documents can all be previewed online. ### Will AI read the files I upload? AI only reads a file when you actively attach it to a conversation or app. Unattached files are never analyzed automatically. Files are semantically indexed to support search, but that is separate from conversation-context injection. ### How do I clean up files in bulk? The Files page supports drag-to-select box selection; once selected you can batch download, delete, or move. You can also set automatic deletion conditions in Cleanup Rules (by type, source, size, path, or age), which run daily at 2 AM. ## Troubleshooting & Support ### What browsers are supported? YeeroAI works best on modern browsers: Chrome, Firefox, Safari, and Edge. Use the latest version for full functionality (such as Service Worker and SSE streaming). ### Does YeeroAI work on mobile? Yes, the web interface is fully responsive and works in mobile browsers. A native mobile app is in development; the desktop app already offers a native experience. ### What if AI generation fails? Try refreshing the page first. If the issue persists, check that your credit balance is sufficient, or switch to another model. Background streaming resumes automatically after a network interruption — no need to resend manually. ### What if a streaming response is interrupted? YeeroAI uses SSE streaming with reconnection support. After the network recovers, it resumes automatically and already-generated content is not lost. If there's no response for a long time, click "Stop" and resend. ### What if cloud sync isn't working? Make sure the relevant toggle is on in Settings → Cloud Sync. Sync uses debounced pushes (about 1 second); after switching devices, wait a moment or refresh to pull the latest data. Some temporary UI state is intentionally excluded from sync. ### How do I report an issue or suggestion? Submit issues, suggestions, or complaints via Account → Feedback and we'll follow up. For refund appeals, privacy, or legal matters, use the same channel and note the request type. > **💬 Didn't find your answer?** > Tell us via Account → Feedback, or browse the full documentation. # ===== ZH ===== --- # 快速开始 *通过三个简单步骤开始使用 YeeroAI——注册、选择模型、开始构建知识。* > **⚡ 一句话摘要** > YeeroAI 免费起步：几分钟完成注册即送 1,000 积分，从 10+ 厂商的 300+ 模型中自由选择，立即开始第一次多模型对话——无需信用卡。 ## 第一步：创建账户 访问 [yeero.ai](https://yeero.ai)，使用以下三种方式之一注册： - **邮箱** — 输入邮箱地址，接收 6 位验证码，登录即可。无需密码。 - **Google** — 点击"使用 Google 继续"一键登录。 - **GitHub** — 点击"使用 GitHub 继续"通过 GitHub 账户认证。 > **💡 欢迎积分** > 注册即可获得 **1,000 免费积分**（30 天有效期）——无需信用卡。这足以充分探索所有功能。 ### 邀请码 如果你有邀请码，在登录页面点击"有邀请码？"并在登录前输入。邀请码可获取额外积分。 ### 测试账户 想在注册前预览网站？在登录页面点击"试用测试账户"快速查看。注意：测试账户功能有限。 ## 第二步：选择模型 YeeroAI 为你提供 **300+ AI 模型**，来自 OpenAI、Anthropic、Google、Meta 等顶级提供商——全部通过 OpenRouter 用一个账户访问。 ### 为对话选择模型 开始新对话时，点击输入区域的 **模型** 按钮打开模型选择器。你可以： - **搜索** — 在搜索栏输入模型名称。 - **按提供商筛选** — 从 OpenAI、Anthropic、Google、Meta、Mistral 等中选择。 - **仅免费** — 切换"仅免费"查看免费模型（每次调用仅消耗 1 积分）。 - **按模态筛选** — 按输入/输出类型筛选：文本、图片、音频、视频、文件。 - **多选** — 选择最多 10 个模型在同一对话中并行比较。 ### 模型市场 从侧边栏访问 **模型** 页面，浏览所有可用模型的详细信息，包括上下文长度、定价、支持的模态和提供商详情。 ## 第三步：开始构建知识 选择模型后，在输入区域输入消息，按 **Enter** 发送（`Shift+Enter` 换行）。你选择的模型将同时响应，你可以并排比较它们的答案。 > **💡 AI 自动命名** > 不用担心为对话命名——AI 会根据对话上下文自动生成标题。告别"新对话 1、2、3..." ## 界面导览 以下是 YeeroAI 界面的快速概览： | 区域 | 描述 | | --- | --- | | 侧边栏（左侧） | 在对话、应用、文件、文件夹、知识、模型和设置之间导航。可使用展开/折叠按钮。 | | 对话列表 | 按文件夹组织的所有对话。搜索、排序和置顶对话。 | | 标签栏 | IDE 风格的标签管理多个打开的对话。右键可关闭、关闭其他、关闭左侧/右侧。 | | 消息区域 | 主要对话视图。滚动浏览消息、查看推理过程和与 AI 响应交互。 | | 输入区域 | 输入消息、选择模型、附加文件、配置参数和发送。 | | 面包屑 | 显示当前文件夹路径。点击任何片段向上导航文件夹层级。 | ## 接下来？ - 了解[对话与交流](https://yeero.ai/docs/zh/chat.html)——多模型对话、消息操作等。 - 探索[分支思维](https://yeero.ai/docs/zh/branching.html)——派生、合并、探索不同的思路。 - 发现[AI 应用](https://yeero.ai/docs/zh/ai-apps.html)——用 AI 生成和迭代 HTML 应用。 - 理解[知识库](https://yeero.ai/docs/zh/knowledge-base.html)——AI 如何从你的对话中提炼知识。 --- # 对话与交流 *YeeroAI 的核心——多模型并行对话、消息操作、流式响应，以及高效 AI 对话所需的一切。* > **⚡ 一句话摘要** > YeeroAI 对话支持最多 10 个模型并行回答、结果并排对比，任何一条消息都可以一键复制、分支、追问或重新生成。 ## 创建对话 有多种方式开始新对话： - **从侧边栏** — 点击对话列表中的"+"按钮或"新建对话"。 - **从文件夹** — 右键点击文件夹，选择"新建对话"在该文件夹内创建。 - **从全局搜索** — 按 `Ctrl`+`K`（Mac 上为 `Cmd`+`K`），选择"新建对话"。 ## 多模型并行对话 YeeroAI 的标志性功能——向多个 AI 模型提出同一问题，并排比较它们的答案。 ### 工作原理 1. 点击输入区域的**模型**按钮。 2. 从模型选择器中选择最多 **10 个模型**。 3. 输入你的消息，按 **Enter** 发送。 4. 每个选中的模型同时生成自己的响应。 5. 响应并排显示，方便比较。 ### 添加更多模型回复 收到响应后，你可以为同一条消息添加更多模型。点击"添加模型回复"选择额外的模型——它们将基于相同的对话上下文生成响应。 ## 发送消息 ### 输入区域 对话底部的输入区域提供多个控件： | 控件 | 描述 | | --- | --- | | 文本输入 | 输入消息。按 **Enter** 发送，`Shift+Enter` 换行。 | | 模型按钮 | 打开模型选择器，选择哪些 AI 模型响应。 | | 工具按钮 | 打开工具选择器，启用 AI 工具（如可用）。 | | 文件夹按钮 | 选择将对话保存在哪个文件夹中。 | | 附件按钮 | 上传文件（图片、文档、音频、视频）随消息一起发送。 | | 设置按钮 | 配置此对话的生成参数。 | ### 文件附件 点击附件按钮或使用拖放将文件附加到消息中： - **支持的格式** — 图片、文档、音频和视频文件。 - **最大文件大小** — 每个文件 512 MB。 - **多文件** — 每条消息可附加最多数量的文件。 - 文件会自动上传并作为 AI 的上下文包含。 > **⚠ 上传进行中** > 文件上传完成前无法发送消息。请等待所有上传完成后再发送。 ## 消息操作 每条 AI 响应消息通过消息工具栏提供多种操作： | 操作 | 描述 | | --- | --- | | **复制文本** | 将消息内容复制为纯文本到剪贴板。 | | **复制为图片** | 将消息复制为图片，用于分享或文档。 | | **保存为文件** | 将消息内容下载为文件（Markdown 格式）。 | | **删除** | 删除消息。此操作无法撤销。 | | **从此消息派生** | 从这条消息创建新分支。参见[分支思维](https://yeero.ai/docs/zh/branching.html)。 | | **设为选定** | 在多模型对话中将此响应标记为"最佳答案"。 | | **添加到文件夹** | 将此消息保存到特定文件夹进行组织。 | 想针对回答中的某段文字深入探讨？直接选中该文字即可发起 [划词追问](https://yeero.ai/docs/zh/followup.html)，在独立线程中层层展开，不影响主对话。 ## 推理过程 某些 AI 模型（如带扩展思考的 Claude）会提供推理过程。当可用时： - 点击**"查看推理"**展开模型的思考过程。 - 点击**"隐藏推理"**折叠。 - 推理部分显示模型在得出答案之前的逐步思考过程。 ## 知识摘要 AI 可以从每对消息中生成关键知识摘要。点击**"查看摘要"**查看提炼的知识。这会反馈到你的[知识库](https://yeero.ai/docs/zh/knowledge-base.html)。 ## 流式响应 AI 响应使用服务器发送事件（SSE）实时流式传输。你会看到响应逐字生成。响应状态实时显示： | 状态 | 含义 | | --- | --- | | 等待中 | 请求已排队。 | | 准备中 | 模型正在初始化。 | | 推理中 | 模型正在思考（推理模型）。 | | 流式传输 | 内容正在生成并流式传输。 | | 已完成 | 响应已完全生成。 | | 错误 | 出了问题。检查错误消息。 | | 已取消 | 生成被手动停止。 | ### 停止生成 点击**停止生成**按钮中止正在进行的 AI 响应。你将保留到那一刻为止已生成的内容。 ### 重新生成 点击**重新生成**让模型对同一输入生成新的响应。 ## AI 工具 YeeroAI 内置丰富的 AI 工具，让模型不仅能对话，还能主动获取信息、执行计算和操作设备能力。工具分为三类，可在对话中同时使用、并行执行。 ### 联网与信息检索 | 工具 | 功能 | | --- | --- | | **联网搜索** | AI 实时搜索互联网，获取最新信息回答你的问题。 | | **网页抓取** | 从指定网址提取网页正文内容。 | | **学术搜索** | 搜索学术论文，获取前沿研究成果。 | ### 计算与时间 | 工具 | 功能 | | --- | --- | | **计算器** | 高精度数学表达式计算。 | | **当前时间** | 获取任意时区的当前日期和时间。 | | **时区转换** | 在不同时区之间转换时间。 | ### 代码执行 | 工具 | 功能 | | --- | --- | | **Python 执行器** | 在安全沙箱中运行 Python 代码，适合数据处理和自动化任务。 | | **JavaScript 执行器** | 在安全沙箱中运行 JavaScript 代码。 | ### 文件与数据 | 工具 | 功能 | | --- | --- | | **文档解析** | 从 PDF、Word、Excel、PPT 文件中提取文本内容。 | | **HTTP 接口测试** | 发送 HTTP 请求并返回响应结果，适合调试接口。 | | **股票数据** | 获取实时股票行情和金融数据。 | | **天气查询** | 查询任意地点的当前天气和预报。 | | **地图与地理编码** | 将地址转换为坐标并搜索附近地点。 | ### 设备能力（前端工具） 以下工具需要访问你的设备能力，调用前会弹窗请求你的许可： | 工具 | 功能 | | --- | --- | | **获取位置** | 获取你当前的地理位置信息。 | | **语音输入** | 通过麦克风录制语音并转换为文字。 | | **读取文件** | 读取你选择的本地文件内容。 | ### 如何启用工具 在输入区域点击**工具按钮**打开工具选择器，勾选需要的工具即可。AI 会根据对话内容自动判断何时调用工具。你也可以在文件夹或对话设置中预设默认启用的工具。 > **💡 并行执行** > 多个工具可同时执行，无需等待前一个完成。例如 AI 可以同时查询天气和获取你的位置，再综合结果给出建议。 ## 消息导航大纲 在聊天界面右侧，导航大纲以可折叠面板的形式展示当前对话的标题结构，帮助你快速定位长对话中的关键内容。 - **展开 / 收起** — 点击大纲面板的展开/收起按钮切换显示状态。 - **一键跳转** — 点击大纲中的任意标题，消息区域自动滚动到对应位置。 - **滚动同步高亮** — 滚动消息时，大纲中对应的标题会自动高亮。 - **全局显示偏好** — 在[设置](https://yeero.ai/docs/zh/settings.html)中选择默认显示模式：始终展开、始终折叠或不展示。移动端自动隐藏大纲。 ## 后台流式传输 当在设置中启用**后台流式传输**时，即使你导航离开对话或刷新页面，AI 生成也会继续。当你返回对话时，你会收到在你离开期间生成的任何内容。 > **💡 弹性连接** > 后台流式传输会在网络中断后自动恢复。你不会丢失任何已生成的内容。 ## 标签页管理 YeeroAI 使用 IDE 风格的标签系统管理多个打开的对话： - **打开对话** — 点击侧边栏中的任何对话在新标签中打开。 - **切换标签** — 点击标签切换到该对话。 - **关闭标签** — 点击标签上的 ×，或右键查看更多选项。 ### 标签上下文菜单 右键点击任何标签可访问： - **刷新消息** — 从服务器重新加载对话。 - **重命名** — 更改对话标题。 - **关闭** — 关闭此标签。 - **关闭其他** — 关闭除当前标签外的所有标签。 - **关闭全部** — 关闭所有打开的标签。 - **关闭左侧全部** — 关闭此标签左侧的所有标签。 - **关闭右侧全部** — 关闭此标签右侧的所有标签。 ## 输入历史 YeeroAI 自动保存你发送的消息，方便复用： - **全部标签** — 显示所有对话的输入历史。 - **当前标签** — 仅显示当前活跃对话的输入历史。 - 点击任何历史条目将其插入输入区域。 - **删除** — 移除单个历史条目。 - **清除全部 / 清除当前** — 清除所有对话或当前对话的历史。 ## 对话管理 ### 从侧边栏 右键点击侧边栏中的任何对话可访问管理选项： - **置顶 / 取消置顶** — 将对话置顶到列表顶部以便快速访问。 - **编辑信息** — 重命名对话。 - **移动** — 将对话移动到其他文件夹。 - **删除** — 永久删除对话。此操作无法撤销。 - **分支** — 查看、检出、重命名、派生、合并、强制执行或删除分支。 - **启用/禁用记忆** — 切换此对话的 AI 记忆。 - **清除记忆设置** — 重置记忆设置为继承父文件夹。 - **清除未引用消息** — 移除未引用的消息以清理对话。 ### 对话设置 每个对话可以有覆盖文件夹和全局默认值的自定义设置： - **模型** — 为此对话选择默认模型。 - **工具** — 启用或禁用 AI 工具。 - **启用记忆** — 允许 AI 记住关键信息。 - **上下文摘要** — 使用摘要而非完整内容以减少 token 使用。 - **生成参数** — 微调温度、采样等。 ### 排序和筛选 在对话列表侧边栏中，你可以： - **搜索** — 按名称搜索文件夹和对话。 - **排序** — 按"最近添加"（默认）或"最早添加"排序。 - **新建文件夹 / 新建对话** — 在当前位置创建新项目。 ## 批量操作 对话页支持鼠标拖拽矩形框选，快速圈选多条内容后进行批量操作。 ### 操作方式 1. 在消息列表空白区域按住鼠标左键开始拖拽，屏幕上出现拖拽矩形框选框。 2. 松开鼠标后，框选范围内的内容自动高亮选中。 3. 页面顶部浮层显示已选数量和可用操作按鈕。 4. 点击所需操作按鈕，批量执行。 ### 支持的列表 | 列表 | 可用批量操作 | | --- | --- | | 消息列表 | 复制文本、复制为图片、保存为 Markdown、添加到文件夹、删除 | | 消息附件弹窗文件列表 | 仅支持框选高亮，暂无批量操作按钮 | > **💡 快捷退出** > 按 `Esc` 随时退出多选模式，恢复普通浏览状态。 ## 积分追踪 每个 AI 响应显示该特定生成消耗的**积分数**。这为你提供使用情况的实时透明度。访问[计费与积分](https://yeero.ai/docs/zh/billing.html)了解详细的积分管理。 ## 生成参数 微调 AI 模型如何生成响应。通过输入区域的设置按钮访问参数： | 参数 | 描述 | | --- | --- | | **最大 Token 数** | 响应中的最大 token 数。留空使用模型默认值。 | | **上下文窗口大小** | 对话中保留的历史消息数量。 | | **推理力度** | 模型思考的力度：无、最低、低、中、高、极高。 | | **温度** | 控制随机性。较低=更专注，较高=更有创意。 | | **响应多样性** | 核采样 (top_p)。控制词语选择的多样性。 | | **词汇多样性** | 频率惩罚。减少相同词汇的重复。 | | **话题发散度** | 存在惩罚。鼓励模型谈论新话题。 | | **摘要上下文** | 使用消息摘要而非完整内容，减少 token 使用。 | ### 图片生成参数 - **图片尺寸** — 1K、2K 或 4K 分辨率。 - **宽高比** — 支持 1:1、2:3、3:2、3:4、4:3、4:5、5:4、9:16、16:9、21:9。 ### 音频生成参数 - **语音** — Alloy、Echo、Fable、Onyx、Nova 或 Shimmer。 - **音频格式** — WAV、MP3、FLAC、Opus 或 PCM16。 ### 视频生成参数 - **视频比例** — 16:9、9:16、1:1、4:3、3:4、21:9、9:21。 - **分辨率** — 480p、720p、1080p、1K、2K、4K。 - **时长** — 视频长度（秒）。 - **生成音频** — 在生成的视频中包含音频。 --- # 分支思维 *用 Git 风格的分支探索每一条思路。派生对话、合并分支，再也不会丢失任何思路。* > **⚡ 一句话摘要** > 分支思考为每个对话提供 Git 式分支：派生探索不同方向、合并保留最佳思路、随时回溯到任意节点——主线对话永远不受影响。 ## 什么是分支？ 在 YeeroAI 中，每个对话默认有一个**主分支**。分支功能让你可以从任何节点创建替代的对话路径，类似于 Git 分支。这对于以下场景非常有用： - **探索替代方案** — 尝试不同的提示词或方法，而不影响主对话。 - **比较策略** — 并行运行实验并比较结果。 - **保留上下文** — 在发送有风险的提示之前派生，这样你可以随时回到原始状态。 - **组织主题** — 在同一对话的不同分支中保持不同子主题。 ## 派生分支 有多种方式创建新分支： ### 从消息派生 将鼠标悬停在任何 AI 响应消息上，点击**"从此消息派生"**。这会创建一个从该消息上下文开始的新分支。 ### 从侧边栏派生 右键点击侧边栏中的对话，选择**"分支"**，然后点击**"派生"**。输入新分支的名称并确认。 ### 从节点派生 使用消息节点的上下文菜单，选择**"从此节点派生"**，创建一个基于到该节点为止的对话历史的新分支。 > **💡 分支命名** > 给分支起描述性的名称，这样以后可以轻松识别。分支名称最长可达最大字符限制。 ## 切换分支 要在对话中切换到不同的分支： 1. 右键点击侧边栏中的对话，选择**"分支"**。 2. 你将看到所有分支及其消息数量的列表。 3. 点击目标分支上的**"检出"**。 4. 消息区域更新为所选分支的对话内容。 当前活跃的分支标记为**"当前"**。**主分支**始终可用。 ## 合并分支 合并会将源分支的消息追加到当前（目标）分支： 1. 切换到你想要合并到的分支（目标分支）。 2. 右键点击对话，选择**"分支"**，然后点击**"合并分支"**。 3. 选择你想要合并的**源分支**。 4. 查看警告：源分支的消息将被追加到当前分支。 5. 点击**"确认合并"**。 > **⚠ 合并仅追加** > 合并会将源分支的消息追加到当前分支的末尾。它不会交错或重写消息。 ## 强制执行分支 强制执行会将当前分支强制指向与源分支相同的位置。当你想要重置分支以匹配另一个分支时很有用： 1. 切换到你想要更新的分支（目标分支）。 2. 右键点击对话，选择**"分支"**，然后点击**"强制执行分支"**。 3. 选择你想要强制执行的源分支。 4. 查看警告并点击**"确认强制执行"**。 > **⚠ 谨慎使用** > 强制执行后，当前分支将指向源分支的位置。当前分支上超出该位置的任何消息可能会变为未引用状态。 ## 重命名分支 要重命名分支： 1. 右键点击侧边栏中的对话，选择**"分支"**。 2. 点击你想要重命名的分支上的**"重命名"**。 3. 输入新名称并确认。 > **💡 唯一名称** > 分支名称在同一对话中必须唯一。如果你尝试使用已存在的名称，将会收到错误提示。 ## 分支过期策略 你可以配置自动清理积累了过多旧消息的分支： 1. 右键点击对话，选择**"分支"**，然后点击某分支上的**"过期策略"**。 2. 开启**"启用过期策略"**。 3. 设置**过期时长**（以月为单位）。 4. 超过指定时长的消息将被自动清理。 ## 删除分支 要删除分支： 1. 右键点击对话，选择**"分支"**。 2. 点击你想要删除的分支上的**"删除"**。 3. 查看警告——该分支中的所有消息将被永久删除。 4. 点击**"确认删除"**。 > **⚠ 无法撤销** > 删除分支是永久性的。分支中的所有消息将被移除且无法恢复。主分支不能被删除。 ## 分支限制 | 限制项 | 值 | | --- | --- | | 每个对话的最大分支数 | 100 | | 每个对话的最大消息数 | 10,000 | | 主分支 | 不可删除 | --- # 划词追问 *读到 AI 回复里的某句话想深入？选中它，一键发起追问。追问在独立线程中展开，支持层层深入、原文虚线标注定位，主对话的上下文保持完整不受影响。* > **⚡ 一句话摘要** > 追问功能允许你选中 AI 回复中的任意一句话，在独立线程中逐层深入探讨——深度挖掘不会打断主对话的节奏。 ## 概览 传统做法里，追问意味着把新问题直接发到主对话里，一来上下文被稀释，二来回头找不到当时讨论的细节。划词追问把每次追问收敛成一条独立线程： - **以选中文字为锚点**——每条追问线程都记录你选中的原文片段，随时可以回到上下文 - **多层深入**——追问的回答还能继续追问，一层层深入，面板会标注当前位于"第 N 层" - **不打断主对话**——主对话的消息流保持原样，追问线程独立保存与管理 - **流式回答**——追问的回答同样逐字流式输出，可随时停止或重新生成 > **📚 与分支思维的区别** > [分支思维](https://yeero.ai/docs/zh/branching.html)面向"同一问题换种思路重答"，会生成可切换的对话分支；追问面向"针对回答中的某个细节继续探讨"，以选中文字为起点层层展开。两者可以配合使用。 ## 发起追问 1. 在 AI 回复中**选中任意一段文字**，选区上方出现浮动工具条 2. 点击"**追问**"——以选中文字为引用创建新的追问线程；也可以点击"**解释这段**"，让 AI 直接解释选中的内容（解释指令的提示词前缀可在设置中自定义） 3. 追问面板打开，顶部以引用卡片展示你选中的原文，在输入框中写下问题并发送 对话输入框旁还有一个追问入口按钮，显示当前对话的追问线程数量，点击即可打开线程列表。 ## 追问面板 追问面板由三部分组成： - **引用卡片**——展示发起追问时选中的文字，点"返回原文"可回到原文所在位置；也可以在这里删除整条线程 - **消息列表**——追问的问答以流式方式呈现，每条回答支持复制、删除、停止生成与重新生成 - **输入框**——继续向当前线程提问 面板头部显示层级指示（如"第 2 层"）与面包屑导航；同一段原文创建了多个线程时，可通过下拉菜单在同级线程之间切换；面板支持展开 / 收起 / 关闭。 ## 多层深入 追问的力量在于层层深入：在追问面板里，**AI 的回答同样可以选中文字**——选中后再次发起追问，就进入下一层（面板层级指示随之 +1）。每一层都会在当前层的回答文本上留下一个虚线标注，标记被追问的位置；点击虚线标注即可跳转到对应的追问线程。 配合面包屑与"返回线程列表"，你可以在任意层级之间自由往返，像探索树一样梳理一个复杂问题。 ## 线程管理 - **线程列表**——当前对话的全部追问线程集中展示，每条线程显示选中文本、消息数量与最近活跃时间 - **删除线程**——删除前二次确认，删除将永久移除线程中的所有消息 - **导出**——一键把整条线程的问答内容复制到剪贴板，方便粘贴到文档或分享 - **更多分享方式**——复制为图片、保存为文件、添加到文件夹 ## 模型与参数 追问回答使用的模型与参数继承自当前对话标签页：你在对话中选择哪个模型，追问就用哪个模型回答，无需额外配置；切换对话标签页后发起的追问，自然跟随新标签页的模型设置。 ## 延伸阅读 - [AI 对话](https://yeero.ai/docs/zh/chat.html)——追问功能依托的对话体系 - [分支思维](https://yeero.ai/docs/zh/branching.html)——同一问题的多路径探索 - [提示词模板](https://yeero.ai/docs/zh/prompt-templates.html)——自定义"解释这段"的提示词前缀 --- # AI 应用 *创建、迭代和讨论 AI 驱动的 HTML 应用——从想法到可用原型只需几秒，支持完整的版本管理和实时预览。* > **⚡ 一句话摘要** > YeeroAI AI 应用把自然语言变成可运行的 HTML 或 Python 应用：补丁式增量更新、完整版本历史、实时预览与内置讨论协作。 ## 什么是 AI 应用？ YeeroAI 中的 AI 应用是由 AI 生成的自包含 HTML 应用。用自然语言描述你的想法，AI 将生成一个功能完整的网页应用。然后你可以通过对话进行迭代、查看实时预览，并管理版本。 ## 创建应用 要创建新的 AI 应用： 1. 从侧边栏导航到**应用**页面。 2. 点击**"新建应用"**。 3. 填写创建表单： **应用名称** — 给应用起一个描述性的名称（如"JSON 格式化工具"）。 4. **描述**（可选）— 简要描述应用的功能。 5. **保存到文件夹** — 选择一个文件夹来组织你的应用，或留在根目录。 6. **模型** — 选择用于生成应用的 AI 模型。 ## 应用详情页 点击任何应用打开其详情页，它提供三个主要视图： ### 预览 **预览**标签页显示应用的实时渲染。HTML 内容显示在自动同步当前主题（浅色/深色）的 iframe 中。如果尚未生成内容，在对话中发送消息即可生成。 ### 代码 **代码**标签页显示当前版本的原始 HTML 源代码。你可以： - **复制代码** — 将整个 HTML 源代码复制到剪贴板。 - **下载源码** — 将 HTML 文件下载到你的电脑。 - **保存更改** — 手动编辑代码并保存为新版本。你必须提供更改摘要。 > **⚠ 代码大小限制** > 应用代码不能超过 500,000 个字符。如果内容过大，你将无法提交手动编辑。 ### 讨论 **讨论**标签页是专门用于规划和迭代应用的对话空间： - **先规划再构建** — 描述你的应用想法，AI 将帮助澄清需求并规划方案。 - **从规划生成** — 点击**"从规划生成应用"**，基于讨论内容创建应用。 - **包含代码** — 开启"包含代码"让 AI 在讨论过程中看到当前应用代码，从而提供更有上下文的建议。 ## 应用对话 每个应用都有自己的对话用于迭代设计。输入区域提供： | 控件 | 描述 | | --- | --- | | 文本输入 | 描述更改或新功能（如"在右上角添加深色模式切换按钮"）。 | | 模型按钮 | 选择用于生成响应的 AI 模型。 | | 工具按钮 | 启用 AI 工具。 | | 设置按钮 | 配置生成参数。 | | 附件按钮 | 附加文件作为参考。 | 每次 AI 响应都会创建应用的一个新**版本**，你可以预览、回滚或继续迭代。 ## 后台生成 AI 应用的内容生成与讨论支持**后台续跑** — 即使你关闭应用详情页、刷新浏览器，甚至临时切走网络，已发起的生成任务都会在后台继续进行，重新打开时会自动衔接到最新进度。 - **关页面继续跑** — 关闭详情页后任务不会被取消，AI 在后台默默把活干完。 - **断线自动恢复** — 网络抖动期间错过的生成内容，重连后会一次性补齐，不会缺帧。 - **多端状态一致** — 同一账号下所有页面都能感知到正在跑的任务，刷新后 UI 状态与后台真实状态完全同步。 - **列表实时角标** — 应用列表会在卡片右上角显示“后台生成中”提示，鼠标悬停查看说明，多任务并行也一目了然。 - **取消即同步** — 在任何端取消生成，其他端会同步停止接收，状态不打架。 > **⚙ 适用范围与开启方式** > 该能力同时覆盖应用的**内容生成**（生成 HTML 输出）与**讨论流**（与应用多轮对话）。需要在**设置 → 后台推流**中手动开启，开启后对话与 AI 应用共用同一开关，同时生效。 ## 版本管理 每次 AI 生成或手动编辑都会创建一个新版本： - **版本列表** — 查看所有版本及其编号（v1、v2、v3...）。 - **当前版本** — 活跃版本标记为"当前"。 - **回滚** — 在任何版本上点击"回滚到此版本"将其设为活跃版本。当前版本不会被删除——仅改变活跃指针。 - **积分追踪** — 每个版本显示消耗的积分和生成时间。 ## 运行应用 点击任何应用上的**"运行"**在浮动对话框中打开它： - **最小化** — 将运行中的应用最小化到底部的紧凑栏。 - **还原** — 将最小化的应用恢复到完整视图。 - **关闭** — 关闭运行中的应用对话框。 - **编辑** — 直接跳转到应用的编辑页面。 - **在应用页面打开** — 在完整详情页中打开应用。 - **导航** — 使用"上一个/下一个"按钮在应用之间浏览。 ## 应用管理 ### 从应用列表 在应用页面，你可以： - **搜索** — 按名称或描述搜索应用。 - **按状态筛选** — 全部、活跃、已归档或草稿。 - **排序** — 最近更新、最多运行或最早创建。 ### 应用操作 点击任何应用卡片上的菜单可访问： - **运行** — 启动应用。 - **编辑应用** — 打开应用详情页。 - **编辑信息** — 更改应用名称和描述。 - **归档 / 取消归档** — 将应用移入或移出归档。 - **删除** — 永久删除应用及其所有版本和聊天记录。 - **移动到文件夹** — 将应用移动到其他文件夹。 - **重命名** — 更改应用名称。 - **置顶 / 取消置顶** — 将应用置顶到列表顶部。 - **启用 / 禁用云同步** — 切换应用运行时数据的云端同步。 ## 云同步 应用运行时数据可以同步到云端： - **启用云同步** — 应用数据在你的设备间同步。 - **禁用云同步** — 应用数据仅保留在本地。 > **⚠ 存储限制** > 每个应用的云端同步数据限制为 **1 MB**。如果数据超出此限制，同步将被暂停。你需要清除应用数据才能恢复同步。 ## 应用状态 | 状态 | 描述 | | --- | --- | | 草稿 | 应用已创建但尚未生成内容。 | | 活跃 | 应用已生成内容且功能完整。 | | 已归档 | 应用已被归档，在默认视图中隐藏。 | ## 批量操作 应用页支持鼠标拖拽矩形框选，快速圈选多个应用或讨论消息后进行批量操作。 ### 操作方式 1. 在列表空白区域按住鼠标左键开始拖拽，屏幕上出现矩形框选框。 2. 松开鼠标后，框选范围内的内容自动高亮选中。 3. 页面顶部浮层显示已选数量和可用操作按鈕，点击执行。 ### 支持的列表 | 列表 | 可用批量操作 | | --- | --- | | 应用列表 | 批量删除、批量归档、批量取消归档、批量移动到文件夹 | | 讨论消息列表 | 批量删除 | > **💡 快捷退出** > 按 `Esc` 随时退出多选模式。 --- # 知识库 *AI 自动从你的对话中提取关键知识，构建向量索引，并在未来对话中注入相关上下文——你的知识库会随着每次对话变得更聪明。* > **⚡ 一句话摘要** > YeeroAI 自动从对话中提炼知识、构建向量索引，并在后续对话中注入最相关的上下文——无需任何手动记录，知识库越聊越聪明。 ## 什么是知识库？ YeeroAI 的知识库是一个自动策划的洞察集合，从你的对话中提取而来。每次你与 AI 对话时，系统会识别关键信息并创建可搜索和引用的知识条目。 这创造了一个**自我改进的知识系统**——你聊得越多，知识库就越聪明，AI 在未来对话中提供相关上下文的能力就越强。 ## 工作原理 当你与 AI 进行对话时，系统会： 1. **监控对话** — 跟踪你与 AI 模型之间的消息交换。 2. **提取知识** — 从消息对中识别关键事实、见解和信息。 3. **构建向量索引** — 为语义搜索创建嵌入向量。 4. **注入上下文** — 当记忆功能启用时，相关知识会自动作为上下文包含在未来的对话中。 > **💡 自动过程** > 知识提取是自动发生的。你不需要做任何事情——只需正常聊天，系统会处理其余的工作。 ## 查看知识 从侧边栏导航到**知识**页面，浏览所有知识条目。每个条目显示： - **标题** — 摘要标题（点击可编辑）。 - **内容** — 提炼的知识内容。 - **标签** — 自动生成的分类标签。 - **原始对话链接** — 点击"查看原始对话"跳转到来源对话。 ## 管理知识 ### 搜索 使用页面顶部的搜索栏搜索知识摘要。 ### 按标签筛选 点击任何标签来筛选知识条目。使用"清除"移除标签筛选。 ### 排序 - **最近更新**（默认）— 最新的条目优先。 - **最早创建** — 最旧的条目优先。 ### 编辑标题 点击任何知识标题即可内联编辑。按 Enter 保存。 ### 归档 / 取消归档 点击"归档"将条目从主视图中隐藏。切换到"已归档"标签页查看和取消归档条目。 ### 添加到文件夹 将知识条目保存到文件夹以便更好地组织。 ## 知识操作 每个知识条目提供以下操作： | 操作 | 描述 | | --- | --- | | 复制内容 | 将知识文本复制到剪贴板。 | | 复制为图片 | 将知识复制为图片。 | | 保存为文件 | 将知识内容下载为 Markdown 文件。 | | 添加到文件夹 | 保存到文件夹进行组织。 | | 归档 / 取消归档 | 隐藏或显示条目。 | | 查看原始对话 | 导航到来源对话。 | ## 记忆与知识 知识和记忆协同工作，使你的 AI 对话更智能： - **记忆** — 当为对话或文件夹启用时，AI 在生成响应时会使用相关的知识条目作为上下文。 - **上下文摘要** — 启用后，系统使用摘要知识而非完整消息内容，大幅减少 token 使用量，同时保留关键信息。 参见[设置](https://yeero.ai/docs/zh/settings.html)了解如何在全局、文件夹或对话级别启用记忆和上下文摘要。 ## 批量操作 知识库列表支持鼠标拖拽矩形框选，快速圈选多条知识后进行批量操作。 在列表空白区域按住鼠标左键拖拽生成框选框，松开后范围内内容自动高亮，页面顶部浮层展示可用操作：**归档**、**取消归档**、**批量添加到文件夹**。按 `Esc` 退出多选模式。 --- # 搜索 *一个强大的搜索界面，覆盖一切——Ctrl/Cmd+K 全局搜索、AI 驱动的语义搜索、分类筛选等等。* > **⚡ 一句话摘要** > 一个 Ctrl/Cmd+K 搜遍对话、应用、文件、知识与记忆：AI 语义搜索 + 分类筛选 + 最近访问 + 搜索历史，全部集中在同一界面。 ## 搜索概述 YeeroAI 提供统一的搜索体验，让你可以在整个平台中查找任何内容。从任何地方按 `Ctrl`+`K`（Mac 上为 `Cmd`+`K`）打开搜索对话框。 ## 全局搜索 通过 `Ctrl`+`K` 访问的全局搜索对话框提供： - **快速搜索** — 输入关键词搜索对话、消息、文件、文件夹、模型、设置和应用。 - **回车搜索** — 按 Enter 执行搜索。 - **方向键导航** — 使用 ↑↓ 导航结果。 - **AI 驱动** — 语义搜索理解你查询背后的含义，而不仅仅是关键词。 ## 搜索分类 结果按分类组织，方便筛选： | 分类 | 包含内容 | | --- | --- | | 全部 | 同时搜索所有分类。 | | 标签页 | 当前打开的对话标签。 | | 对话 | 你的对话列表。 | | 消息 | 对话中的单条消息，包括带有相似度分数的语义搜索结果。 | | 文件夹 | 对话、消息和应用文件夹。 | | 模型 | 可用的 AI 模型及提供商信息。 | | 设置 | 应用设置和偏好。 | | 我的文件 | 用户上传的文件。 | | AI 文件 | AI 生成的文件。 | | 应用 | 你的 AI 应用程序。 | ## 「全部」分类 不确定内容在哪个分类时，选择「全部」即可一次并发搜索所有分类，结果按分类分组展示，无需逐一切换查找。 - **并发搜索** — 选中「全部」后，对话、消息、文件、应用、模型等分类同时开始搜索。 - **分组展示** — 结果按分类清晰分组，清晰明了不混乱。 - **语义搜索生效** — AI 语义搜索在「全部」分类下同样生效，跨分类匹配语义相关内容。 > **💡 何时用「全部」** > 当你记得某段内容却不确定它是对话、消息还是文件时，「全部」分类是最好的选择。如果你已确定内容类型，切到具体分类可以获得更词准的结果。 ## AI 语义搜索 对于消息和文件，YeeroAI 支持基于向量嵌入的语义搜索： - 用自然语言描述你要找的内容。 - AI 基于**含义**查找结果，而不仅仅是关键词匹配。 - 结果包含**相似度分数**（百分比），显示与查询的匹配程度。 > **💡 语义 vs 关键词** > 当你知道你要找什么但不知道确切的词时，使用语义搜索。当你知道具体术语时，使用关键词搜索。 ## 搜索历史 YeeroAI 会记住你最近的搜索： - 之前的搜索查询会自动保存。 - 点击历史条目重新运行搜索。 - **删除** — 移除单个历史条目。 - **清除历史** — 移除所有搜索历史。 - 历史条目显示相对时间戳（刚刚、几分钟前、几小时前、几天前）。 ## 最近访问 搜索对话框还显示最近访问的项目： - 你最近访问过的对话、消息、文件、模型、设置、文件夹和应用。 - **移除** — 从最近访问中移除单个项目。 - **清除全部** — 清除整个最近访问列表。 ## 快捷操作 当你打开搜索对话框且未输入关键词时，会展示一组快捷操作卡片，帮你快速完成常用操作。 ### 操作类型 | 类型 | 说明 | 示例 | | --- | --- | --- | | 页面导航 | 快速跳转到平台各页面 | 对话、应用、知识库、文件、设置等 | | 快速创建 | 一键新建内容 | 新建对话、新建文件夹、新建应用、上传文件 | | 命令执行 | 快速执行常用命令 | 切换主题、切换语言、查看积分、邀请有礼 | | 设置跳转 | 直达常用设置项 | 提示词、记忆、默认模型、云同步 | ### 智能排序 快捷操作会根据你的使用频率自动排序，常用操作自动置顶。你也可以点击“重置”恢复默认排序。 > **💡 搜索过滤** > 在搜索框中输入关键词时，快捷操作也会根据关键词自动过滤，帮你快速找到目标操作。 ## 结果操作 从搜索结果中，你可以： - 点击任何结果直接导航到它。 - **前往对话** — 打开消息的来源对话。 - **添加到文件夹** — 将消息保存到文件夹。 - **删除** — 移除消息（需确认）。 - **下载文件** — 下载文件结果。 - 对于对话，选择要查看的特定分支。 --- # CLI 快速入门 *三步上手 yeero 命令行工具——安装、登录、开始对话。* > **⚡ 一句话摘要** > yeero 命令行工具一条命令安装、2 分钟上手：安装、登录，然后在终端里直接与任意模型对话。 ## 什么是 yeero CLI？ yeero CLI 是 YeeroAI 的命令行客户端，通过在本地运行一个后台 daemon 服务与云端 AI 平台通信。你可以用它： - 在终端发起多轮 AI 对话（TUI 或行式模式） - 管理 AI 生成的 Python 应用（创建、运行、版本管理） - 浏览对话历史、模型列表和可用工具 - 自动化脚本与管道场景（`--oneshot` 模式） > **💡 与网页端/桌面端共享账号** > CLI 登录后与网页端、桌面端共用同一账号，对话历史实时同步。 ## 第一步：安装 ### macOS / Linux 在终端运行一行命令，脚本会自动检测平台、下载最新版并配置 PATH： `curl -fsSL https://yeero.ai/cli/install.sh | sh` 安装完成后，脚本自动将 yeero 路径写入 `~/.bashrc`、`~/.zshrc` 或 `~/.config/fish/config.fish`（取决于你的 shell）。新终端窗口即可直接使用。 如需安装指定版本： `YEERO_VERSION=0.1.0 curl -fsSL https://yeero.ai/cli/install.sh | sh` Windows（PowerShell） 以普通用户权限打开 PowerShell，运行： `irm https://yeero.ai/cli/install.ps1 | iex` 脚本自动将 yeero 安装目录（`%LOCALAPPDATA%\yeero`）写入用户级 PATH，当前会话立即生效，新终端窗口同样可用。 如需安装指定版本： `$env:YEERO_VERSION="0.1.0"; irm https://yeero.ai/cli/install.ps1 | iex` 验证安装 `yeero --version` 输出版本号即表示安装成功。 ### 安装环境变量 | 变量 | 默认值 | 说明 | | --- | --- | --- | | `YEERO_VERSION` | 最新版 | 指定要安装的版本号 | | `YEERO_INSTALL_DIR` | `~/.local/bin`（macOS/Linux） `%LOCALAPPDATA%\yeero`（Windows） | 自定义安装目录 | ## 第二步：登录 首次使用前需要登录 YeeroAI 账号（使用邮箱验证码，无需密码）： `yeero login` CLI 会引导你输入邮箱，发送 6 位验证码后再输入验证码完成登录。也可以直接带参数跳过交互： `yeero login --email your@email.com # 收到验证码后 yeero login --email your@email.com --code 123456` 💡 登录仅需一次Token 安全存储在系统 Keychain 中，重启后无需重新登录。 验证登录状态： `yeero whoami` 第三步：发起对话 登录后直接运行 `yeero chat` 即可进入 TUI 模式（全屏交互界面），开始与 AI 对话： `yeero chat` 也可以直接带上消息内容： `yeero chat "帮我写一个快速排序算法"` 指定模型（支持多个）： `yeero chat -m claude-3-5-sonnet "分析这段代码的复杂度"` 一次性模式（适合脚本/管道）： `yeero chat --oneshot "今天天气怎么样？" echo "写一个 hello world" | yeero chat --plain --oneshot` 💡 自动启动 Daemon执行任何需要网络的命令时，CLI 会自动检查并启动后台 daemon，无需手动运行 `yeero start`。 ## 接下来？ - 查看[命令参考](https://yeero.ai/docs/zh/cli-commands.html)——所有命令的完整语法与参数 - 深入了解[对话详解](https://yeero.ai/docs/zh/cli-chat.html)——TUI 模式、多媒体生成、采样参数 - 探索[AI 应用](https://yeero.ai/docs/zh/cli-app.html)——用 CLI 管理 Python AI 应用 - 了解[配置与 Daemon](https://yeero.ai/docs/zh/cli-config.html)——语言、默认模型、守护进程管理 --- # CLI 命令参考 *yeero CLI 所有子命令的语法、参数和示例。* > **⚡ 一句话摘要** > 本页是 yeero CLI 全部子命令的完整参考：语法、选项与可直接复制的示例，一站查全。 ## 命令总览 | 命令 | 说明 | | --- | --- | | yeero start | 启动后台 daemon 服务 | | yeero stop | 停止 daemon 或所有 yeero 进程 | | yeero restart | 重启 daemon | | yeero status | 查看 daemon 运行状态 | | yeero login | 邮箱验证码登录 | | yeero logout | 退出登录 | | yeero whoami | 查看当前登录账号 | | yeero chat | 发起 AI 对话（TUI / 行式 / 一次性） | | yeero chats | 列出对话与文件夹 | | yeero models | 列出可用 AI 模型 | | yeero tools | 列出可用工具 | | yeero app | 管理 Python AI 应用 | | yeero do | 用自然语言意图执行 Python AI 应用 | | yeero config | 管理本地配置 | | yeero update | 检查并自动更新 CLI | ## 全局选项 以下选项适用于所有子命令： | 选项 | 说明 | | --- | --- | | `--lang ` | 临时覆盖界面语言（`zh` 或 `en`），不写入配置 | | `--log-level ` | 日志级别（`trace`｜`debug`｜`info`｜`warn`｜`error`） | | `-h, --help` | 输出帮助信息 | | `-V, --version` | 输出版本号 | ## yeero start 启动后台 daemon HTTP 服务。大多数命令会自动按需启动 daemon，通常无需手动运行。 `yeero start [OPTIONS]` 选项默认说明 `-p, --port ``3000`监听端口 `--host ``127.0.0.1`监听地址 `-d, --detach`—后台运行，spawn 子进程后立即返回 `# 前台启动（日志输出到终端） yeero start # 后台运行 yeero start --detach # 指定端口 yeero start --port 8080 --detach` yeero stop 停止 daemon 服务。 `yeero stop [OPTIONS]` 选项说明 `-f, --force`强制结束，不等待优雅关闭 `--all`停止所有 yeero 进程（含 chat / app run / TUI 等，隐含强杀） `yeero stop yeero stop --force yeero stop --all` yeero status 查看 daemon 是否在运行及监听端口。 `yeero status` yeero restart 重启后台 daemon：先停止正在运行的 daemon，再以 detached 模式在后台重新拉起。若 daemon 原本未运行，则直接启动。 `yeero restart [OPTIONS]` 选项说明 `-f, --force`强制结束旧进程，不等待优雅关闭 `# 优雅重启 yeero restart # 强制重启 yeero restart --force` yeero login 邮箱验证码登录。CLI 不支持 OAuth，如需使用 Google / GitHub 登录请使用网页端或桌面端。 `yeero login [OPTIONS]` 选项说明 `--email `预填邮箱，跳过交互输入 `--code `预填验证码，跳过发码步骤（需同时传 `--email`） `# 交互式登录 yeero login # 非交互登录（适合 CI/脚本） yeero login --email you@example.com --code 123456` yeero logout 退出当前账号。默认只退出本设备并清除本地 Token；使用 `--all` 可登出全部设备。 `yeero logout [OPTIONS]` 选项说明 `--all`登出该账号在所有设备上的登录状态 `yeero logout yeero logout --all` yeero whoami 打印当前登录账号信息（JSON 格式）。未登录时输出提示。 `yeero whoami` yeero chat 发起 AI 对话。支持 TUI 全屏模式、行式多轮模式和一次性模式。详细使用见[对话详解](https://yeero.ai/docs/zh/cli-chat.html)。 `yeero chat [MESSAGE] [OPTIONS]` 常用参数： | 选项 | 说明 | | --- | --- | | `MESSAGE` | 初始消息（可选，不传则进入空白会话） | | `--plain` | 使用行式模式（替代 TUI） | | `--oneshot` | 发送消息后立即退出，适合脚本/管道 | | `-m, --model ` | 指定模型 ID，可重复传入（多模型并行） | | `--chat-id ` | 接续现有会话 | | `--branch-id ` | 接续现有分支 | | `--folder-id ` | 新建对话时指定所属文件夹 | | `--tool ` | 启用工具（可重复传入） | | `--user-file ` | 上传本地文件作为附件（可重复传入） | | `--use-knowledge` | 启用知识库检索 | `yeero chat "解释什么是向量数据库" yeero chat -m gpt-4o -m claude-3-5-sonnet "对比两个模型的回答" yeero chat --oneshot "今天日期是？"` yeero chats 列出当前账号的对话与文件夹。 `yeero chats [OPTIONS]` 选项默认说明 `--folder-id ``0`（根目录）指定父文件夹 ID `--name `—按名称模糊搜索 `--pinned`—仅返回置顶项 `--order ``DEFAULT``DEFAULT` 或 `EARLIEST` `--page ``1`页码 `--size ``50`每页条数（1~200） `--json`—以 JSON 格式输出（脚本友好） `yeero chats yeero chats --name "架构" --json` yeero models 列出当前账号可用的 AI 模型。 `yeero models [OPTIONS]` 选项说明 `--keyword `按名称模糊搜索 `--free`仅展示免费模型 `--input-modalities `按输入能力过滤，逗号分隔（如 `text,image`） `--output-modalities `按输出能力过滤，逗号分隔（如 `text,audio`） `--page / --size`分页参数 `--json`以 JSON 格式输出 `yeero models yeero models --keyword claude --free yeero models --output-modalities image --json` yeero tools 列出当前账号可用的工具。 `yeero tools [OPTIONS]` 选项说明 `--keyword `按 ID / 名称 / 描述模糊搜索 `--json`以 JSON 格式输出 `yeero tools yeero tools --keyword search --json` yeero app 管理 Python AI 应用。包含多个子命令，详细使用见[AI 应用](https://yeero.ai/docs/zh/cli-app.html)。 | 子命令 | 说明 | | --- | --- | | `yeero app create` | 创建新 Python 应用 | | `yeero app list` | 列出所有应用 | | `yeero app show` | 查看应用详情 | | `yeero app save` | 保存本地文件为新版本 | | `yeero app versions` | 列出版本历史 | | `yeero app version-show` | 查看指定版本详情 | | `yeero app rollback` | 回滚到指定版本 | | `yeero app generate` | AI 代码生成（单轮） | | `yeero app discuss` | AI 多轮讨论 REPL | | `yeero app run` | 运行应用 | | `yeero app stop` | 停止运行中的应用 | | `yeero app uninstall` | 卸载本地已安装应用 | ## yeero do 用自然语言描述意图，自动路由并执行最合适的 Python AI 应用。详细使用见[AI 应用执行](https://yeero.ai/docs/zh/cli-do.html)。 `yeero do [INTENT] [OPTIONS]` 选项说明 `INTENT`意图描述（自然语言）。省略时提示输入。 `--dry-run`仅预演路由结果，不执行。 `--name `未匹配到应用时，为新脚本指定名称。 `--no-create`未匹配到应用时不创建新脚本。 `--model `创建新应用时用于生成代码的模型 ID。 yeero config 管理本地配置。详细使用见[配置与 Daemon](https://yeero.ai/docs/zh/cli-config.html)。 | 子命令 | 说明 | | --- | --- | | `yeero config get-lang` | 查看当前界面语言 | | `yeero config set-lang ` | 设置界面语言（`zh` 或 `en`） | | `yeero config get-default-model` | 查看默认模型 | | `yeero config set-default-model ` | 设置默认模型 | | `yeero config unset-default-model` | 清除默认模型设置 | | `yeero config add-to-path` | 将 yeero 目录加入系统 PATH | | `yeero config get-path` | 查看 PATH 中已有的 yeero 条目 | | `yeero config clear-path` | 从 PATH 移除所有 yeero 条目 | ## yeero update 检查并自动更新 CLI 到最新版本。 `yeero update [OPTIONS]` 选项说明 `--check`仅检查是否有新版本，不下载安装 `# 检查更新并自动安装 yeero update # 仅检查，不安装 yeero update --check` 💡 自动检查更新每次运行 CLI 命令时，会在后台静默检查（1 小时 TTL 缓存），有新版本时显示一行黄色提示，不阻塞当前命令执行。 --- # CLI 对话详解 *深入了解 `yeero chat`——行式多轮模式、一次性模式、多媒体生成与采样参数。* > **⚡ 一句话摘要** > yeero chat 支持多轮与一次性模式、多模型并行回答、文件与工具输入、知识库上下文增强，以及完整的采样参数控制。 ## 两种交互模式 ### 行式多轮模式（默认） `yeero chat` 默认进入行式多轮模式：在终端提示符后输入消息，按 Enter 发送，AI 回复流式输出；可持续多轮对话，输入 `/exit` 或按 Ctrl+C 退出。 `# 启动行式多轮（新对话） yeero chat # 携带首轮消息启动 yeero chat "分析一下这段代码" # 接续已有对话 yeero chat --chat-id 12345` 一次性模式 加上 `--oneshot` 发送一条消息后立即退出，适合脚本、管道和 CI 场景： `# 直接输出 AI 回复 yeero chat --oneshot "用一句话总结量子计算" # 管道输入 cat error.log | yeero chat --plain --oneshot "帮我解释这个错误" # 关闭颜色输出（适合管道或日志采集） yeero chat --no-color --oneshot "生成一段 JSON"` 多模型并行 通过多次传入 `-m` 同时向多个模型发送相同消息，并排比较回答： `yeero chat -m gpt-4o -m claude-3-5-sonnet "哪个编程语言更适合 AI 开发？" yeero chat -m deepseek-r1 -m gemini-2.5-pro "解这道数学题：..."` 不传 `-m` 时使用默认模型（可通过 `yeero config set-default-model` 设置）。 ## 会话与分支管理 | 选项 | 说明 | | --- | --- | | `--chat-id ` | 接续指定会话（从已有历史继续） | | `--branch-id ` | 接续指定分支 | | `--folder-id ` | 新建对话时指定所属文件夹 | `# 查找对话 ID yeero chats --name "上次的项目" # 接续该对话 yeero chat --chat-id 98765 "继续之前的讨论"` 文件与工具 附件上传 使用 `--user-file` 上传本地文件作为对话附件（可重复传入多个文件）： `yeero chat --user-file ./report.pdf "总结这份报告的要点" yeero chat --user-file image1.png --user-file image2.png "比较这两张图片"` 启用工具 使用 `--tool` 指定工具 ID（可通过 `yeero tools` 查看可用工具）： `yeero tools # 列出可用工具 yeero chat --tool web_search "今天有什么科技新闻？"` 知识库检索 `yeero chat --use-knowledge "根据我的历史对话，总结我关注的主要技术领域"` 采样参数 精细控制模型生成行为： | 选项 | 范围 | 说明 | | --- | --- | --- | | `--temperature` | 0~2 | 采样温度，越高越随机（默认由模型决定） | | `--nucleus-sampling` | 0~2 | Top-P 文本多样性 | | `--maximum-token-count` | 正整数 | 最大生成词元数量 | | `--presence-penalty` | 0~2 | 话题发散度，减少重复话题 | | `--frequency-penalty` | 0~2 | 词汇丰富度，减少重复词汇 | | `--reasoning-effort` | XHIGH｜HIGH｜MEDIUM｜LOW｜MINIMAL｜NONE | 推理程度（支持推理的模型） | | `--message-window-size` | 0–100 | 上下文历史消息窗口大小 | `yeero chat --temperature 0.2 --reasoning-effort HIGH "写一篇严谨的技术分析" yeero chat --temperature 1.5 "给我一个创意十足的故事开头"` 图片生成 选择支持图片输出的模型（如 dall-e-3、gpt-image-1），并传入图片参数： | 选项 | 可选值 | 说明 | | --- | --- | --- | | `--image-aspect-ratio` | RATIO_1_1｜RATIO_2_3｜RATIO_3_2｜RATIO_4_3｜RATIO_3_4｜RATIO_16_9｜RATIO_9_16 | 图片宽高比 | | `--image-size` | SIZE_1K｜SIZE_2K｜SIZE_4K | 图片分辨率 | `yeero chat -m dall-e-3 --image-aspect-ratio RATIO_16_9 --image-size SIZE_2K "一幅赛博朋克城市夜景"` 音频生成 选择支持音频输出的模型（如 gpt-4o-audio），传入音频参数： | 选项 | 可选值 | 说明 | | --- | --- | --- | | `--audio-voice` | ALLOY｜ECHO｜FABLE｜ONYX｜NOVA｜SHIMMER | 合成语音音色 | | `--audio-format` | WAV｜MP3｜FLAC｜OPUS｜PCM16 | 音频输出格式 | `yeero chat -m gpt-4o-audio --audio-voice NOVA --audio-format MP3 "朗读这段文字：..."` 视频生成 选择支持视频输出的模型，传入视频参数（视频生成为异步任务，CLI 会自动轮询结果）： | 选项 | 可选值 | 说明 | | --- | --- | --- | | `--video-aspect-ratio` | RATIO_16_9｜RATIO_9_16｜RATIO_1_1｜RATIO_4_3｜RATIO_3_4 | 视频宽高比 | | `--video-resolution` | P480｜P720｜P1080｜K1｜K2｜K4 | 视频分辨率 | | `--video-duration` | 1–300（秒） | 视频时长 | | `--video-generate-audio` | true｜false | 是否同时生成音频轨道 | ``` yeero chat -m wan-2.1 --video-aspect-ratio RATIO_16_9 --video-resolution P720 --video-duration 10 "一只猫在草地上奔跑" ``` --- # CLI AI 应用 *用 `yeero app` 管理 Python AI 应用的完整生命周期——从创建到运行，从版本管理到回滚。* > **⚡ 一句话摘要** > yeero app 在终端中管理 Python AI 应用的完整生命周期：创建、生成代码、运行、版本管理与回滚。 ## 概览 yeero 的 AI 应用是由 AI 生成和迭代的 Python 脚本，在本地 Python 虚拟环境中运行。CLI 提供完整的管理能力： | 子命令 | 说明 | | --- | --- | | `app create` | 创建新应用（云端注册 + 本地初始化） | | `app list` | 列出所有应用及状态 | | `app show` | 查看应用详情 | | `app save` | 保存本地文件为新版本 | | `app versions` | 列出版本历史 | | `app version-show` | 查看指定版本详情与代码 | | `app rollback` | 回滚到指定版本 | | `app generate` | AI 单轮生成/修改代码 | | `app discuss` | AI 多轮讨论 REPL | | `app run` | 在本地运行应用 | | `app stop` | 停止运行中的应用 | | `app uninstall` | 卸载本地应用（清除脚本和 venv） | ## 创建应用 `yeero app create [OPTIONS]` 选项说明 `--name `应用名称（不传则交互输入） `--description `应用描述 `--folder `所属文件夹名称 `yeero app create --name "数据分析助手" --description "分析 CSV 数据并生成报表"` 列表与详情 `# 列出所有应用 yeero app list # 按状态过滤（DRAFT、ACTIVE 等） yeero app list --statuses ACTIVE # 按名称搜索 yeero app list --keywords "分析" # JSON 格式输出 yeero app list --json # 查看应用详情 yeero app show my-app yeero app show --json` AI 代码生成 `app generate` 向 AI 发送单轮请求，直接生成或修改应用代码： `yeero app generate my-app --prompt "添加数据可视化功能，生成折线图" # 指定编辑模式 yeero app generate my-app --prompt "优化性能" --mode patch # 指定推理强度 yeero app generate my-app --prompt "重构代码结构" --effort high` 选项默认说明 `--prompt `交互输入生成指令 `--mode``auto`编辑模式：`auto`｜`full`｜`patch` `--effort``medium`推理强度：`minimal`｜`low`｜`medium`｜`high`｜`xhigh` `--model `应用配置指定模型 `--include-current``true`是否将当前代码附入上下文 `--tool `—启用工具（可重复） `--user-file `—附件文件（可重复） AI 多轮讨论 `app discuss` 进入多轮 REPL，与 AI 持续对话来迭代应用代码： `yeero app discuss my-app # 指定模型和推理强度 yeero app discuss my-app --model claude-3-5-sonnet --effort high` 在 REPL 中输入需求，AI 会根据当前代码上下文给出修改建议或直接生成代码。输入 `/exit` 或 `Ctrl+C` 退出。 ## 运行与停止 `# 运行应用 yeero app run my-app # 带初始输入运行（JSON 字符串） yeero app run my-app --input '{"query": "分析最近30天的销售数据"}' # 停止运行中的应用 yeero app stop ` 💡 自动安装依赖首次运行时，CLI 会自动创建独立的 Python 虚拟环境并安装应用所需依赖。 ## 版本管理 ``` # 保存当前本地文件为新版本 yeero app save my-app --file ./app.py --change-summary "添加了数据可视化功能" # 从 stdin 读入代码 cat new_version.py | yeero app save my-app --file - --change-summary "重构" # 列出版本历史 yeero app versions my-app # 查看指定版本详情 yeero app version-show ``` ## 回滚 ``` # 交互式选择版本回滚（带确认提示） yeero app rollback my-app # 指定版本 ID 回滚 yeero app rollback my-app --version # 跳过确认直接回滚 yeero app rollback my-app --version -y ``` ## 卸载 ``` # 卸载单个应用（清除本地脚本和 venv） yeero app uninstall my-app # 跳过确认 yeero app uninstall my-app -y # 卸载所有本地已安装应用 yeero app uninstall --all -y ``` --- # AI 应用执行 *用一句话描述你想做的事，yeero 自动匹配或创建 Python 应用并执行。不用记各平台命令，不用每次烧 Token。* > **⚡ 一句话摘要** > yeero do 把一句话变成行动：自动匹配或创建合适的 Python 应用并本地执行——运行不耗积分，也无需记忆任何命令。 ## 功能概述 `yeero do` 的本质仍然是执行 `yeero app` 中的 Python 应用脚本。区别在于你不需要记住应用 ID 或子命令，只需用一句话描述意图，系统会： 1. 在本地已同步的 Python App 中搜索最匹配的一个； 2. 如果匹配度高，直接执行该应用； 3. 如果找不到合适应用（或匹配度较低），会先创建一个由 AI 生成代码的新 Python App，再执行它。 因此，`yeero do` 既适合"我想运行某个已有应用"的场景，也适合"我有个新想法，直接帮我生成并跑起来"的场景。 另一个典型场景：**再也不用记各平台的命令了**。Windows 用 `ipconfig`、Linux / macOS 用 `ifconfig` 或 `ip addr`；不同系统查看进程、清理缓存、管理服务的命令各不相同——现在只需一句 `yeero do "查看本机 IP 地址"`，不用操心当前是什么操作系统。 更重要的是，`yeero do` **极其节省 Token 费用**。与每次对话都要调用大模型的 AI Agent 不同，`yeero do` 仅在首次创建应用时消耗一次 LLM 调用；之后每次执行都是直接运行本地 Python 脚本，不再产生任何 Token 开销。高频使用的工具类任务，用 `yeero do` 代替 Agent 可以帮你省下可观的费用。 ## 语法与参数 `yeero do [INTENT] [OPTIONS]` 参数 / 选项说明 `INTENT`意图描述（自然语言）。省略时会提示输入。 `--dry-run`仅预演路由结果，不真正执行。 `--name `未匹配到应用时，为新创建的脚本指定名称；默认自动生成。 `--no-create`未匹配到应用时不创建新脚本，直接失败。 `--model `创建新应用时用于生成代码的模型 ID；不传则使用应用默认模型。 执行流程 当你运行 `yeero do "..."` 时，CLI 会先发一次 dry-run 探测： - **高匹配**：若存在明显匹配的应用，直接执行；若有多个高分候选，会弹出候选列表让你用上下键选择（非终端环境会回退到输入序号）。 - **未匹配 / 弱匹配**：会提示确认用于生成新应用的模型，确认后开始创建、安装并执行。 执行过程中，`creating`、`generating`、`installing`、`running` 等阶段提示会实时输出到终端。 ## 示例 ``` # 我想做某事，不用管怎么做——yeero 自动匹配或创建应用 yeero do "把这 200 张照片按拍摄日期分到文件夹里" # 跨平台命令不用记——不用查 Windows 还是 Linux 该用什么 yeero do "查看哪个进程占了最多内存" # 日常自动化——一句话搞定重复性工作 yeero do "把这个 CSV 里重复的行去掉，按金额排序后导出 Excel" # 只看看会匹配到哪个应用，不真正执行 yeero do "生成一个番茄钟网页" --dry-run # 新应用指定名称和生成模型 yeero do "写一个定时备份数据库并上传 OSS 的脚本" --name db-backup --model gpt-4o # 匹配不到就直接失败，不创建新应用 yeero do "随便试试" --no-create ``` ## 与 yeero app run 的区别 | | `yeero do` | `yeero app run` | | --- | --- | --- | | 输入方式 | 自然语言意图 | 应用 ID / 名称 | | 匹配逻辑 | 自动路由，必要时创建新应用 | 直接运行指定应用 | | 适用场景 | “我想做某事” | “我要运行这个应用” | | 本质 | 最终执行的仍是 `yeero app` 的 Python 脚本 | ## 注意事项 - `yeero do` 需要 daemon 运行；未运行时会自动按需启动。 - 首次创建新应用需要等待 AI 生成代码、安装依赖，耗时可能较长。 - 如果生成的应用运行失败，CLI 会提示应用 ID，可用 `yeero app discuss ` 继续迭代修复。 --- # 配置与 Daemon *使用 `yeero config` 管理语言、默认模型及 PATH 环境，使用 `yeero start / stop / status` 控制后台 Daemon 服务。* > **⚡ 一句话摘要** > yeero config 管理语言、默认模型与 PATH 配置；yeero start / stop / status 控制支撑 CLI 运行的后台守护进程。 ## config 命令概览 `yeero config` 包含以下子命令，均为纯本地文件操作，无需 Daemon 运行： | 子命令 | 说明 | | --- | --- | | `config get-lang` | 查看当前界面语言 | | `config set-lang` | 设置界面语言 | | `config get-default-model` | 查看当前默认模型 | | `config set-default-model` | 设置默认模型 | | `config unset-default-model` | 清除默认模型（恢复内置默认） | | `config add-to-path` | 将 yeero 加入系统 PATH | | `config get-path` | 查看 PATH 中已有的 yeero 条目 | | `config clear-path` | 从 PATH 移除所有 yeero 条目 | ## 语言设置 yeero CLI 支持中文（`zh`）和英文（`en`）两种界面语言。 `# 查看当前语言 yeero config get-lang # 设置为中文 yeero config set-lang zh # 设置为英文 yeero config set-lang en # 不传参数将进入交互式输入 yeero config set-lang` 语言配置持久化保存在 `config.toml` 中，路径按构建档位隔离（开发 / 预发 / 生产分别独立）。 > **💡 全局 --lang 选项** > 每条命令也可通过 `--lang zh` / `--lang en` 临时覆盖语言，优先级高于配置文件。 ## 默认模型 默认模型决定了 `yeero chat`、`yeero app generate` 等命令在未指定 `--model` 时所使用的模型。 `# 查看当前默认模型 yeero config get-default-model # 设置默认模型 yeero config set-default-model gpt-4o # 不传参数将进入交互式输入 yeero config set-default-model # 清除默认模型（回退到内置默认值） yeero config unset-default-model` 模型 ID 可通过 `yeero models` 命令获取完整列表。 ## 将 yeero 加入 PATH 安装后如果终端无法识别 `yeero` 命令，可运行以下命令自动配置 PATH： `# 添加到用户级 PATH（推荐） yeero config add-to-path # 添加到系统级 PATH（需要管理员权限） yeero config add-to-path --system # 预览操作但不实际执行 yeero config add-to-path --dry-run # 从 PATH 移除 yeero config add-to-path --remove` 选项说明 `--system`写入系统级 PATH（默认写入用户级） `--remove`从 PATH 移除 yeero 所在目录 `--dry-run`仅预览，不实际修改 💡 macOS / Linux此命令会在 `~/.bashrc`、`~/.zshrc` 等 Shell 配置文件中追加 PATH 导出语句，修改后需重开终端或执行 `source ~/.zshrc` 生效。 ## 查看 PATH 配置 使用 `config get-path` 可以查看当前 yeero 可执行文件所在目录，以及用户级和系统级 PATH 中已加入的 yeero 条目： `yeero config get-path` 输出示例： `当前 yeero 目录: /usr/local/bin 用户级 PATH 中的 yeero 条目: /usr/local/bin 系统级 PATH: 未包含 yeero 条目` 从 PATH 移除 yeero `config clear-path` 用于从 PATH 中移除所有 yeero 相关条目（不追加新路径），适合卸载或迁移场景： `# 从用户级 PATH 移除（默认） yeero config clear-path # 从系统级 PATH 移除（需要管理员权限） yeero config clear-path --system # 预览操作但不实际执行 yeero config clear-path --dry-run` 选项说明 `--system`从系统级 PATH 移除（默认从用户级） `--dry-run`仅预览，不实际修改 💡 与 add-to-path --remove 的区别`clear-path` 与 `add-to-path --remove` 效果完全一致，是更语义化的别名，便于直觉理解。 ## Daemon 概览 yeero CLI 的大部分功能（对话、AI 应用等）依赖一个后台 Daemon 进程。Daemon 以本地 HTTP 服务形式运行，负责与云端 API 通信、管理对话上下文和 AI 应用生命周期。 通常情况下，运行任意需要网络的命令时 Daemon 会自动启动。你也可以手动管理它： ## 启动 Daemon `# 启动 Daemon（后台运行） yeero start # 指定监听端口（默认自动分配） yeero start --port 7700 # 前台运行（调试用途） yeero start --foreground` 停止 Daemon `yeero stop` 此命令会向正在运行的 Daemon 发送终止信号，优雅退出。 ## 查看状态 `yeero status` 输出示例： `Daemon running on port 7700 Version: 0.1.0` 若 Daemon 未运行，将显示 `Daemon not running`。 ## 配置文件路径 yeero 的配置文件 `config.toml` 按构建档位隔离存放： | 系统 | 路径 | | --- | --- | | macOS / Linux | `~/.config/yeero/config.toml` | | Windows | `%APPDATA%\yeero\config.toml` | 配置文件为 TOML 格式，目前支持以下字段： `# 界面语言：zh 或 en lang = "zh" # 默认模型 ID（可选） default_model = "gpt-4o"` ⚠ 注意请勿手动编辑 config.toml 中的其他字段，建议始终通过 `yeero config` 命令修改配置。 --- # 桌面端 *Yeero Desktop 是基于 Tauri 构建的原生桌面应用，提供与网页端相同的全功能 AI 对话体验，并带来原生级别的性能与系统集成能力。* > **⚡ 一句话摘要** > Yeero Desktop 是基于 Tauri 的原生桌面应用，支持 Windows、macOS 与 Linux：完整的 YeeroAI 体验，原生级性能，外加全局启动器、截图录屏与内置编辑器。 ## 产品简介 Yeero Desktop 将 YeeroAI 的完整功能封装为原生安装包，支持 macOS、Windows 和 Linux。与网页端相比，桌面端额外提供： - 无需打开浏览器，随时通过系统 Dock / 任务栏启动 - 原生窗口标题栏，支持明暗主题跟随系统自动切换 - 应用内自动更新，新版本发布后一键安装 - OS 系统通知（含更新提醒） ## 安装 下载与当前系统匹配的安装包： | 系统 | 架构 | 下载 | | --- | --- | --- | | macOS | Apple Silicon (M 系列) | [Yeero-latest-aarch64.dmg](https://yeero.ai/desktop/latest/Yeero-latest-aarch64.dmg) | | macOS | Intel | [Yeero-latest-x64.dmg](https://yeero.ai/desktop/latest/Yeero-latest-x64.dmg) | | Windows | x64 | [Yeero-latest-x64.exe](https://yeero.ai/desktop/latest/Yeero-latest-x64.exe) | | Windows | ARM64 | [Yeero-latest-arm64.exe](https://yeero.ai/desktop/latest/Yeero-latest-arm64.exe) | | Linux | x64 | [Yeero-latest-amd64.AppImage](https://yeero.ai/desktop/latest/Yeero-latest-amd64.AppImage) | | Linux | x64 (deb) | [Yeero-latest-amd64.deb](https://yeero.ai/desktop/latest/Yeero-latest-amd64.deb) | ### macOS 安装步骤 1. 下载对应架构的 `.dmg` 文件 2. 打开 DMG，将 **Yeero** 拖入 Applications 文件夹 3. 首次启动时，若弹出"无法验证开发者"提示，可右键单击 Yeero.app → **打开**，在确认弹窗中选择"打开"即可；或前往 **系统设置 → 隐私与安全性**，点击"仍要打开" > **💡 如何判断 Mac 架构** > 点击左上角苹果菜单 → **关于本机**，芯片显示 M 开头的为 Apple Silicon，显示 Intel 的为 x64。 > **⚠ 提示"应用已损坏，无法打开"** > 若 macOS 弹出"Yeero 已损坏，无法打开"，请打开终端执行以下命令，然后再重新打开应用： `xattr -cr /Applications/Yeero.app` ### Windows 安装步骤 1. 下载 `.exe` 安装程序 2. 运行安装程序，如弹出 SmartScreen 警告，点击"更多信息" → "仍要运行" 3. 按向导完成安装，Yeero 将出现在开始菜单和桌面快捷方式 ### Linux 安装步骤 **AppImage（推荐，免安装）：** `# 赋予执行权限 chmod +x Yeero-latest-amd64.AppImage # 直接运行 ./Yeero-latest-amd64.AppImage` **deb 包：** `sudo dpkg -i Yeero-latest-amd64.deb` ⚠ Linux 运行依赖Linux 版本依赖 `libwebkit2gtk-4.1` 等系统库。如无法启动，请运行： `sudo apt-get install libwebkit2gtk-4.1-0 libgtk-3-0 libayatana-appindicator3-1` ## 首次启动 安装完成后首次打开 Yeero Desktop： 1. 应用自动启动内置 Daemon 服务（无需手动配置） 2. Daemon 就绪后自动加载主界面 3. 点击右上角头像或"登录"按钮，完成账号登录 4. 登录后即可使用全部功能 ## 主要功能 Yeero Desktop 包含网页端全部功能，并额外提供桌面端独有能力： - **网页端全部功能**——多模型对话、分支思维、知识库、AI 应用、20+ 内置工具、文件库、文件夹、Prompt 模板等 - **Launcher**——Alt+Space 全局命令栏：计算、启动应用、文件搜索、截图、录屏、剪贴板历史、AI 对话 - **截图与录屏**——智能窗口检测、放大镜取色、长截图自动滚动拼接、系统音频与麦克风、摄像头画中画 - **图片编辑器**——独立窗口，9 种标注工具、图层面板、EXIF 元数据查看 - **视频编辑器**——多轨时间轴剪辑、7 种转场、滤镜调色、文字与字幕、AI 智能剪辑 - **磁盘清理**——目录树可视化磁盘占用、30+ 分类自动识别应用、AI 分析未知目录 ## 主题切换 Yeero Desktop 支持三种主题模式，可在设置页面或界面右上角切换： - **浅色**——固定使用浅色主题 - **深色**——固定使用深色主题 - **跟随系统**——自动跟随 macOS / Windows 系统主题设置 主题切换会同步更新原生标题栏颜色，保持整体视觉一致。 ## 语言设置 在设置页面可以将应用界面切换为中文或英文。语言设置与 CLI 独立，不会互相影响。 ## 自动更新 Yeero Desktop 内置自动更新机制： - 每次启动时自动检查新版本（最多每小时检查一次） - 将窗口切换回前台时也会触发检查 - 发现新版本后，应用内弹出提示并同时发送系统通知 - 点击提示中的"立即更新"按钮，下载完成后自动重启安装 所有更新包均经过 Ed25519 数字签名验证，确保安全性。 > **💡 手动检查更新** > 也可在设置页面点击"检查更新"主动触发，不受 1 小时缓存限制。 ## 内置 Daemon 与 CLI Yeero Desktop 内置了完整的 `yeero` CLI，并将 Daemon 作为 sidecar 进程随应用一起运行。应用启动时自动拉起 Daemon，退出时自动关闭，无需用户手动管理。 **安装 Desktop 即可在终端使用 yeero 命令。**Desktop 和 CLI 共用同一套 Daemon 后台服务，因此： - 在 Desktop 中登录后，打开终端直接运行 `yeero chat` 即可使用，无需重新登录 - CLI 的所有对话记录、配置、知识库数据与 Desktop 完全同步 - 无需单独安装 CLI 工具链，Desktop 安装包已包含完整的 `yeero` 可执行文件 > **💡 如何在终端使用 yeero 命令** > Desktop 安装完成后，`yeero` 命令会自动注册到系统 PATH（macOS / Linux）或通过系统变量生效（Windows）。打开新的终端窗口，直接运行 `yeero --help` 即可验证。 --- # Launcher *Launcher 是 Yeero Desktop 的全局唤起器：按下 `Alt+Space`，一个输入框即可完成计算、启动应用、搜索文件、截图录屏、粘贴剪贴板历史，甚至直接向 AI 提问——无需离开当前工作。* > **⚡ 一句话摘要** > 启动器是一个全局 Alt+Space 命令栏：计算、启动应用、搜索文件、截图、录屏、剪贴板历史、直接提问 AI——全程不打断当前工作。 > **💡 仅桌面端可用** > Launcher 依赖系统级全局热键与本地能力（应用枚举、窗口控制、文件索引等），仅在 Yeero Desktop 中提供，网页端没有此功能。 ## 概览 Launcher 的设计理念是"一个输入框，解决一切"。你在输入框中键入的任何内容，都会同时交给 16 个处理器并行解析：数字表达式交给计算器、网址交给浏览器、应用名交给应用搜索、自然语言交给 AI……所有结果按相关度合并排序，回车即执行。 - **全局唤起**——默认热键 `Alt+Space`，任何界面下随叫随到；再次按下可隐藏 - **中英双语触发**——所有处理器均支持中文与英文关键词，输入"截图"或 "screenshot" 效果相同 - **拼音与容错**——支持全拼、首字母、同音字与拼写纠错，记不住准确名称也能命中 - **两种呈现方式**——多数结果选中即执行（如打开应用）；部分能力（系统监控、设置、HTML 应用）在 Launcher 窗口内展开面板，执行完不关闭窗口 - **空闲快捷面板**——不输入任何内容时，Launcher 展示常用快捷操作（锁屏、睡眠、静音等），点一下即用 ## 快速上手 1. 按下 `Alt+Space` 唤起 Launcher 2. 直接输入：试试 `12*8`、`截图`、`chrome`、`今天天气怎么样` 3. 用 `↑` `↓` 选择结果，或直接按 `Enter` 执行第一条 输入停顿片刻后，Launcher 会把文本同时分发给全部启用的处理器，结果按匹配度从高到低排列。每个结果都标注了来源处理器与操作说明，悬停可查看快捷键。 ## 匹配规则 Launcher 的匹配引擎针对中英文混合输入做了深度优化： - **模糊匹配**——不必记住完整名称，输入零散的几个字母也能命中，如 `ysbjsq` 也能找到"记事本"之类的应用 - **拼音支持**——全拼（`jisuanqi`）、首字母（`jsq`）均可匹配中文名称 - **同音字**——输入同音别字同样可以命中 - **拼写纠错**——少量字母打错、漏打也不影响，照样有结果 - **多字段评分**——应用搜索会综合显示名、别名、本地化名称、描述等多个字段计分，越接近意图排得越靠前 ## 16 个处理器一览 下表汇总了 Launcher 内置的全部处理器。触发词一栏为典型示例，实际支持更多同义表达。 | 处理器 | 触发示例 | 能力概述 | | --- | --- | --- | | 计算器 | `12*(3+5)` | 实时计算表达式，结果一键复制 | | 网址打开 | `google.com` | 在默认浏览器打开网址 | | 截图 | `截图` / `screenshot` | 全屏选区截图，确认后进入图片编辑器 | | 录屏 | `录屏` / `screen record` | 选区/窗口录制，浮动控制条 | | 系统命令 | `锁屏` / `lock` | 20+ 条系统级命令，危险操作二次确认 | | 系统监控 | `sm` / `cpu` | CPU/内存/磁盘/网络实时面板 | | Yeero 导航 | `设置` / `知识库` | 跳转应用页面、生命周期操作、打开编辑器 | | 应用搜索 | 任意应用名 | 启动已安装应用，拼音/容错匹配 | | 浏览器书签 | `bm github` | 搜索四大浏览器书签并打开 | | 剪贴板历史 | `clip @image` | 检索/恢复历史剪贴板，过滤语法丰富 | | 结束进程 | `kill chrome` | 按名称结束进程（同名分组） | | HTML 应用 | 任意应用名 | 在 Launcher 面板内运行 BROWSER 型 AI 应用 | | 文件搜索 | 任意文件名（≥2 字符） | 本地索引极速搜索，支持深度内容搜索 | | 窗口切换 | 任意窗口标题 | 聚焦已有窗口，含最小化窗口 | | 路由引擎 | 任意自然语言 | AI 识别意图，匹配并执行 AI 应用 | | AI 对话 | 任意问题 | 用默认模型直接提问，流式回答 | ## 处理器详解 ### 计算器 输入纯数字与运算符组成的表达式即触发，结果实时显示在候选列表中，不用按回车等答案。支持 `+` `-` `*` `/` `^`（乘方）`%`（取余）、括号与负数、小数。 - **边输边算**——表达式的计算结果作为第一条候选实时呈现，随时可见 - **回车即复制**——按 `Enter` 把结果复制到剪贴板，转身就能粘贴到文档、表格或聊天窗口 - **结果干净**——`0.1+0.2` 得到干净的 `0.3`，不会出现一长串浮点尾巴 - **空闲快捷入口**——不输入任何内容时，空闲面板也提供计算器入口 示例：`12*(3+5)` → `96`；`2^10` → `1024`；`100%7` → `2`；`-8/4` → `-2`。 ### 网址打开 输入完整链接（`https://…`）或裸域名（如 `github.com`、`yeero.ai`），回车即在系统默认浏览器中打开。 - **自动补全协议**——裸域名自动补上 `https://` 前缀，无需手动输入 - **支持端口与路径**——`localhost:3000/admin`、`192.168.1.1` 等带端口或内网地址同样可用 ### 截图 输入"截图"、"截屏"或 "screenshot"（输入前几个字母即可联想出来）后回车，Launcher 会先隐藏自身，再进入全屏选区，完整流程只需三步： 1. **框选区域**——拖拽框选任意区域，或直接单击某个窗口让系统自动识别窗口边界 2. **确认选区**——确认后自动打开独立的图片编辑器窗口 3. **标注输出**——立即框选重点、涂鸦、加文字，完成后保存到文件库或一键复制到剪贴板 不输入任何内容时，空闲面板也提供截图快捷入口。完整截图流程（含智能窗口识别、长截图拼接）与标注工具见 [截图与录屏](https://yeero.ai/docs/zh/capture.html)。 ### 录屏 输入"录屏"、"屏幕录制"、"screenrecord" 或 "screen record"（空格加不加都行）后回车： 1. **选择录制区域**——拖拽框选任意区域（最小 100×100），或双击某个窗口自动识别 2. **录制中随时控制**——屏幕上会出现浮动控制条，带录制状态指示灯与实时时长，支持暂停、继续与停止 3. **录完即用**——录制完成后，可直接把视频送入视频编辑器剪辑，或在文件管理器中查看文件 - **录制中不打扰**——正在录制时，Launcher 不会再弹出录屏候选项，所有控制都集中在浮动控制条上 - **帧率画质可调**——在 [设置 → Launcher](https://yeero.ai/docs/zh/settings.html) 中按需调整录屏帧率与画质 - **空闲快捷入口**——空闲面板提供录屏入口，不打字也能一键开录 录制还支持麦克风收音与摄像头画中画，完整说明见 [截图与录屏](https://yeero.ai/docs/zh/capture.html)。 ### 系统命令 把常用系统操作变成可搜索的命令，一个关键词直达，再也不用在系统菜单里层层翻找。中英文均可触发，英文连字符可省略（`volumeup` 与 `volume-up` 均可），输入前几个字母也会实时联想： | 命令 | 中文触发 | 英文触发 | | --- | --- | --- | | 锁屏 | 锁屏 | lock | | 睡眠 | 睡眠 | sleep | | 关机 | 关机 | shutdown | | 重启 | 重启 | restart | | 关闭屏幕 | 关闭屏幕 / 息屏 | turn-off-display | | 静音 | 静音 | mute | | 音量加 / 音量减 | 音量加 / 音量减 | volume-up / volume-down | | 播放暂停 | 播放暂停 / 播放 / 暂停 | media-play-pause | | 上一首 / 下一首 | 上一首 / 下一首 | media-previous / media-next | | 表情面板 | 表情 / 表情符号 | emoji-panel | | 任务管理器 | 任务管理器 | task-manager | | 系统设置 | 系统设置 / 设置 | system-settings | | 回收站 / 清空回收站 | 回收站 / 清空回收站 | recycle-bin / empty-recycle-bin | | 显示桌面 | 显示桌面 | show-desktop | | 打开下载 / 打开文档 | 打开下载 / 打开文档 | open-downloads / open-documents | | 用户目录 | 用户目录 / 主目录 / 家目录 | open-home | | 最近文件 | 最近文件 / 最近 | recent / recent-files | 每条命令都支持更多同义表达，如"音量增"、"音量降"、"下载"、"文档"等，怎么顺口怎么输。 - **危险操作有保护**——关机、重启、清空回收站执行前都会弹出确认对话框，不怕手滑 - **最近文件直达**——输入"最近文件"（或 `recent`）会列出系统最近打开的 20 个文件，回车直接用关联程序打开 - **空闲面板常驻**——不输入任何内容时，空闲面板固定展示锁屏、睡眠、显示桌面、关闭屏幕、静音、系统设置、任务管理器七个高频命令，点一下即用 ### 系统监控 输入 `sm`、`monitor`、"系统监控"、"监控"、"系统状态"等关键词，Launcher 会在窗口内展开一个实时系统监控面板，数据持续自动刷新，不用再打开任务管理器。 - **监控内容全面**——CPU 使用率与核心信息、内存与交换分区占用、每块磁盘的容量与使用率、网络上传/下载速率、传感器温度、系统负载、进程数量、系统版本与运行时长，一屏尽览 - **子关键词直达**——输入"CPU"、"内存"、"磁盘"（或 `disk`、"硬盘"、"存储"）会自动打开面板并定位到对应区块 - **面板常驻窗口**——面板在 Launcher 窗口内展开，看完按 `Esc` 关闭即可，不影响手头的工作 - **空闲快捷入口**——空闲面板提供"系统监控"、"CPU"、"内存"、"磁盘"四个快捷入口 ### Yeero 导航 用自然语言在 Yeero 内部跳转与操作，全部支持中英文，还可以在关键词前加"打开"、"前往"、"跳转"等动词（"打开设置"与"设置"效果相同）。命中的内容按三种方式呈现：独立页面在新窗口打开，设置与账号面板直接在 Launcher 窗口内展开，不打断当前操作。 - **打开页面（新窗口）**——"对话"、"新建对话"、"知识库"、"文件夹"、"文件库"、"模型"；说"消息文件夹"/"应用文件夹"会直接打开文件夹页对应分类 - **设置精确到分类**——"设置"打开完整设置面板；"系统设置"、"常规设置"、"提示词设置"、"记忆设置"、"启动器设置"、"文件索引设置"直达对应分类，面板在 Launcher 窗口内展开 - **账号精确到页签**——"账号"、"个人中心"打开账号面板；"积分"、"邀请"、"反馈"直达对应页签 - **生命周期操作**——"主窗口"、"主界面"聚焦主窗口，"退出"退出应用，"重启"重新启动应用 - **打开编辑器**——"图片编辑器"、"修图"、"视频编辑器"、"剪辑"直接打开对应独立窗口 - **帮助文档**——"文档"、"帮助"在系统浏览器打开本文档站，并自动匹配当前语言 - **媒体文件直达**——输入或粘贴图片/视频文件路径，图片用图片编辑器打开，视频用视频编辑器打开。支持引号包裹路径、`file://` 链接，前后加"打开/编辑"等动词也能识别（如 `打开 C:\照片\头像.png`）；直接粘贴一张图片也能立即送入图片编辑器 ### 应用搜索 在 Launcher 中输入任何文字，都会同时在你已安装的应用中搜索——比在开始菜单、启动台里翻找快得多。候选项带应用图标，回车直接启动。 - **怎么输都能搜**——应用名称、拼音全拼（如 `weixin`）、拼音首字母，甚至应用描述里的关键词都能命中 - **中英文互通**——输入"微信"能找到 WeChat，输入 "wechat" 也能找到微信 - **同音字与拼写纠错**——打了同音别字或少打几个字母，依然有结果 - **自动跟上变化**——新安装或卸载应用后，搜索结果会自动更新，无需任何操作 ### 浏览器书签 输入前缀 `bm`、`bookmark` 或"书签"，后跟关键词，即可同时搜索 Chrome、Edge、Firefox、Safari 四大浏览器的书签，回车在默认浏览器中打开。 - **智能排序**——标题完全一致的排最前，其次按标题开头匹配、标题包含、网址包含依次排序，想要的结果总在最上面 - **信息完整**——每条书签显示标题、所在文件夹与网址，一眼看清来源 - **无关键词浏览**——只输入前缀不加关键词，会列出前 50 条书签供直接浏览 - **空闲快捷入口**——空闲面板提供书签快捷入口，不打字也能直接浏览 ### 剪贴板历史 输入前缀 `clip`、`clipboard` 或"剪贴板"，即可检索你复制过的所有内容——文本、图片、文件都会被自动记录。选中历史条目回车，即恢复到剪贴板，可直接粘贴使用，再也不怕刚复制的内容被覆盖。 - **关键词搜索**——前缀后输入任意关键词，按内容模糊查找 - **过滤器语法**——下列过滤词可与关键词任意组合、顺序不限： | 语法 | 含义 | | --- | --- | | `@text`（`@t` / `@txt`） | 仅文本条目 | | `@image`（`@i` / `@img`） | 仅图片条目（带缩略图预览） | | `@file`（`@f`） | 仅文件条目 | | `@tag:工作` | 按标签过滤 | | `@today`（`@td`）/ `@week` / `@month` / `@year` | 今天 / 本周 / 本月 / 今年 | | `@7d` | 最近 N 天（数字可自定义） | | `@date:2026-08-01~2026-08-27` | 指定日期区间 | 例如 `clip @image @week 设计` 表示"本周剪贴板中含'设计'的图片"。 - **条目信息完整**——每条历史显示类型、复制时间、同步状态与置顶/收藏标记；图片条目直接显示缩略图预览 - **跨设备端到端加密同步**——开启云同步后，剪贴板历史会在登录同一账号的多台设备间自动共享；同步内容全程端到端加密，密钥只保存在你自己的设备上，除你之外无人可读。其他设备同步来的加密条目在本机暂时无法解密时会给出提示，在 [设置 → 系统](https://yeero.ai/docs/zh/settings.html) 中配置同一密钥后即可正常查看 - **空闲快捷入口**——空闲面板提供剪贴板历史快捷入口，不打字也能直接浏览最近复制的内容 历史上限、记录开关与跨设备云同步见 [设置 → Launcher](https://yeero.ai/docs/zh/settings.html)；端到端加密密钥在 [设置 → 系统](https://yeero.ai/docs/zh/settings.html) 中配置。 ### 结束进程 应用卡死、无响应时，输入 `kill`、"结束"、"终止"或"杀掉"前缀加进程名，即可快速结束它，无需打开任务管理器层层翻找。 - **中文拼音都能搜**——输入"微信"能命中 Weixin 进程，输入 `xunlei` 能命中"迅雷"；拼写稍有出入也能命中 - **同名进程自动分组**——一个应用往往有多个后台进程，搜索结果会合并为一组，显示实例数量与合计内存占用，回车一次结束整组，干净利落 - **看得懂的名字**——优先显示应用的本地化名称与图标（如"迅雷"而非 Thunder.exe），不必面对陌生的进程文件名 - **失败有提示**——权限不足或进程已退出时会明确说明原因，不会静默失败 ### HTML 应用 你创建的 [AI 应用](https://yeero.ai/docs/zh/ai-apps.html)（BROWSER 型）可以直接在 Launcher 里运行：输入应用名（支持拼音与拼写纠错），命中后选中即在 Launcher 窗口内的面板中打开使用，无需切换到主窗口，用完即走。 - **随叫随到**——每次唤起 Launcher，应用列表都会自动更新，刚创建的应用立即可用 - **面板内完整使用**——应用在面板中完整运行，期间 Launcher 保持打开，方便连续使用 - **没有匹配也有出路**——输入内容没匹配到任何应用时，会提供"在应用中创建"入口，跳转到应用创建流程并自动预填你的输入 ### 文件搜索 输入 2 个及以上字符即参与文件搜索。搜索基于 Yeero 自建的文件索引，响应速度远快于系统搜索，而且越常用的文件排得越靠前。扫描范围与忽略规则在 [设置 → 文件索引](https://yeero.ai/docs/zh/settings.html) 中配置： - **首次使用有向导**——第一次搜索时会引导你完成索引配置，选好扫描范围即可开始 - **边建索引边搜索**——索引扫描进行中，结果区会显示扫描进度，已就绪的部分立即可搜 - **结果信息完整**——每条结果显示文件名（关键词高亮）、所在位置、大小与修改时间，回车用关联程序打开 - **深度内容搜索**——加 `deep:` 前缀（如 `deep:季度报告`）可搜索文件*内容*，基于系统原生索引（Spotlight / Windows Search / Tracker）；该能力默认关闭，需在设置中开启，未开启时会给出引导提示 - **无结果不抓瞎**——搜不到时会提示当前已索引的文件数量，并提供调整扫描范围的入口，帮你判断是索引未覆盖还是确实没有匹配 - **快捷操作**——结果支持 `Alt+Enter` 在文件管理器中显示、`Ctrl/Cmd+Enter` 复制完整路径 ### 窗口切换 输入任意文字匹配已打开窗口的标题、应用名或进程名，回车将窗口恢复并聚焦到前台——避免重复打开同一个应用，窗口再多也能一键直达。 - **覆盖所有窗口**——包括已最小化与收进托盘的窗口，条目上会标注窗口当前状态 - **匹配同样聪明**——拼音、同音字、拼写纠错统统支持，中文窗口标题也能用拼音搜 - **列表始终新鲜**——每次唤起 Launcher 都会重新扫描窗口；若目标窗口刚被关闭，条目会自动消失，不会点到已不存在的窗口 - **平台授权引导**——macOS 首次使用需要授予自动化权限，界面会提供一键跳转系统设置的引导；部分 Linux 环境（如 Wayland）不支持时也会明确说明原因与解决办法 ### 路由引擎 路由引擎让"一句话执行应用"成为可能：直接用大白话描述你想做的事（如"把这段文字翻译成英文"、"统计这份表格的总销售额"），AI 会自动理解你的意图，并在你的 AI 应用中寻找能完成它的脚本。 - **命中即执行**——匹配到应用时，候选项直接展示该应用，回车即运行；执行过程与结果在 Launcher 面板中实时呈现 - **面板内反复调试**——执行完成后，直接在面板中输入新内容回车即可重新运行同一应用，无需从头再来 - **没有匹配也能办**——找不到现成应用时提供两个入口：**"快速创建并运行"**让 AI 现场生成一个新应用并立即执行；**"在应用中创建"**则跳转应用创建页（自动预填你的输入），适合先沟通需求再创建 想深入了解应用体系，见 [AI 应用](https://yeero.ai/docs/zh/ai-apps.html)。 ### AI 对话 任何非空输入都会产生一条"询问 AI"候选——选中后使用你的默认模型直接发起对话，答案在 Launcher 面板中逐字流式输出，可随时停止。 - **输入越具体，排得越靠前**——问题描述得越完整，这条候选越容易排到前列，方便直接回车提问 - **面板内追问**——回答完成后，在面板中输入新问题回车即可继续，上下文连贯 - **适合快速问答**——查一个词、问一个概念、要一段代码片段，不必打开主窗口；需要多模型对比、分支探索或完整对话管理时，请使用主窗口的 [对话功能](https://yeero.ai/docs/zh/chat.html) ## 键盘快捷键 | 快捷键 | 功能 | | --- | --- | | `Alt+Space` | 全局唤起 / 隐藏 Launcher | | `↑` / `↓` | 在候选结果间移动 | | `Enter` | 执行选中项；面板模式执行中按下则作为迭代输入重新执行 | | `Ctrl+1` ~ `Ctrl+9` | 直接执行第 N 条结果 | | `Alt+Enter` | 在文件管理器中显示（文件搜索结果） | | `Ctrl/Cmd+Enter` | 复制文件路径（文件搜索结果） | | `Alt+P` | 置顶 / 取消置顶选中结果 | | `Alt+M` | 打开选中结果的右键菜单 | | `Tab` | 展开 / 收起详情面板 | | `Esc` | 先关闭详情面板，再关闭 Launcher；执行中按下为取消 | ## 相关设置 在 [设置 → Launcher](https://yeero.ai/docs/zh/settings.html) 中可以调整（仅桌面端显示该分类）： - **全局唤起快捷键**——展示当前唤起热键（默认 `Alt+Space`） - **录屏帧率**——5–30，默认 25；越高越流畅，文件越大 - **录屏质量**——H.264 CRF 值 0–51，默认 23，数值越小画质越高 - **剪贴板历史上限**——10–10000 条 - **剪贴板历史**——关闭后将不再记录剪贴板内容 - **剪贴板跨设备同步**——开启后可在多设备间同步剪贴板历史，并提供"立即同步"与"清空云端数据"操作 - **处理器开关**——16 个处理器可逐个启用或禁用，支持按名称与触发词搜索；关闭某个处理器后，Launcher 将不再产生它的结果 > **💡 处理器开关即时生效** > 处理器开关保存在本地并立即生效，无需重启应用。若发现某类结果"消失"了，先来这里检查对应处理器是否被关闭。 ## 延伸阅读 - [截图与录屏](https://yeero.ai/docs/zh/capture.html)——选区、标注、长截图与录制控制的完整流程 - [图片编辑器](https://yeero.ai/docs/zh/image-editor.html)——截图后的标注、图层与元数据工具 - [视频编辑器](https://yeero.ai/docs/zh/video-editor.html)——录制完成后的剪辑、转场与导出 - [AI 应用](https://yeero.ai/docs/zh/ai-apps.html)——HTML 应用与路由引擎背后的应用体系 --- # 截图与录屏 *Yeero Desktop 的屏幕捕获全家桶：截图支持窗口智能识别、放大镜取色与长截图滚动拼接；录屏支持区域 / 全屏 / 窗口识别、系统声音与麦克风、摄像头画中画——捕获完成无缝衔接图片与视频编辑器。* > **⚡ 一句话摘要** > Yeero Desktop 捕捉一切：智能窗口识别截图、RGB 取色放大镜、滚动长截图拼接；录屏支持系统声音、麦克风与摄像头画中画——所有素材直接进入内置编辑器。 > **💡 仅桌面端可用** > 截图与录屏依赖系统级屏幕捕获能力，仅在 Yeero Desktop 中提供。可通过 [Launcher](https://yeero.ai/docs/zh/launcher.html) 输入"截图"或"录屏"快速发起。 ## 截图 ### 选择区域 发起截图后，屏幕进入全屏选区覆盖层，有两种方式圈定范围： - **拖拽框选**——按住鼠标拖出任意矩形区域，选区旁实时显示尺寸（如 1920 × 1080） - **单击窗口**——点击任意窗口，系统自动识别该窗口的边界并贴合选区，无需手动对齐 选区过程中鼠标附近会显示放大镜，除精确对位外还会实时显示当前像素的 **RGB 色值**，取色、对齐一步到位。 ### 确认后的操作 框选完成后出现操作栏： - **复制**——截图直接进入剪贴板，粘贴即用 - **编辑**——打开 [图片编辑器](https://yeero.ai/docs/zh/image-editor.html)，标注后保存或分享 - **长截图**——见下节 - **取消**——放弃本次选区重新选择 ### 长截图 网页、聊天记录这类超出屏幕的内容，用长截图一次抓全。进入长截图模式后有两种捕获方式： - **手动滚动**——你照常滚动鼠标滚轮或触控板，系统在背后逐帧捕获并自动拼接 - **自动滚动**——点击"开始自动滚动"后无需任何操作，系统自动滚动目标窗口并捕获（过程中请勿操作鼠标），可随时停止 捕获过程实时显示帧数与累计高度，达到最大高度自动结束，最终拼接为一张完整长图进入编辑器。 ## 录屏 ### 录制设置 发起录屏后先打开录制设置面板： - **帧率**——可调范围与默认值可在 [设置 → Launcher](https://yeero.ai/docs/zh/settings.html) 中配置，越高越流畅、文件越大 - **质量**——高 / 中 / 低三档（对应不同的编码质量参数） - **格式与输出目录**——自由选择录制文件的保存位置 ### 选择录制范围 - **全屏**——直接录制整个显示器 - **拖拽框选区域**——录制任意矩形区域，最小 100×100，过小会提示"录制区域太小" - **双击窗口**——双击任意窗口自动识别其边界作为录制范围 ### 录制中的控制 开始录制后，屏幕上会出现浮动控制条，实时显示已录制时长，随时可以： - **暂停 / 继续**——中场休息不打断录制，时间轴不留空档 - **停止**——结束本次录制并进入完成面板 ### 音频与摄像头 - **系统音频**——录制电脑播放的声音（会议、视频、提示音） - **麦克风**——录制讲解人声，可选择默认设备之外的输入设备 - **摄像头**——开启后以画中画形式叠加录制画面，可调整位置，适合"真人出镜"的教程与演示 ### 录制完成 停止后弹出完成面板，展示生成的视频分段，可以： - **在编辑器中打开**——直接进入 [视频编辑器](https://yeero.ai/docs/zh/video-editor.html)剪辑 - **打开文件夹 / 复制路径 / 下载**——快速找到录制文件 - **导出分段**——将分段导出为成品文件 两点可靠性说明：首次录制会自动下载视频处理工具（ffmpeg），只需等待一次；如果上一次录制因意外中断，下次启动录屏时会自动尝试修复并提示结果。 ## 延伸阅读 - [图片编辑器](https://yeero.ai/docs/zh/image-editor.html)——截图后的标注与分享 - [视频编辑器](https://yeero.ai/docs/zh/video-editor.html)——录制后的剪辑与导出 - [Launcher](https://yeero.ai/docs/zh/launcher.html)——发起截图与录屏的最快入口 --- # 图片编辑器 *Yeero Desktop 内置的轻量图片编辑器：截图完成后直接进入，也可以打开任意本地图片，用矩形、箭头、马赛克等 9 种工具标注，叠加裁剪、旋转、缩放与背景智能透明等图像处理，配合图层管理与 EXIF 信息查看，覆盖日常修图与问题反馈场景。* > **⚡ 一句话摘要** > 内置图片编辑器覆盖日常标注需求：矩形、箭头、马赛克等 9 种工具，配图层管理与 EXIF 查看器——截图可直接进入编辑。 ## 概览 图片编辑器以独立窗口运行，与主窗口互不干扰，可以同时打开多个实例。围绕"快速标注 + 快速分享"设计： - 支持 PNG、JPG、JPEG、WebP、BMP、GIF 格式 - 9 种标注工具，全部带数字快捷键，选中即用 - 裁剪、旋转、翻转、缩放与背景智能透明等图像处理能力 - 完整的撤销/重做历史，任何操作都可以反悔 - 图层面板管理所有标注对象：显隐、排序、删除 - 一键复制到剪贴板，直接粘贴到聊天工具或文档 - 内置图片信息面板，可查看基本参数与完整 EXIF 数据 ## 打开图片 共有四种方式进入编辑器： 1. **截图后自动打开**——通过 [截图](https://yeero.ai/docs/zh/capture.html) 或 [Launcher](https://yeero.ai/docs/zh/launcher.html) 完成选区确认后，编辑器窗口自动打开并载入截图 2. **Launcher 打开文件**——在 Launcher 中粘贴或输入图片路径，回车即用编辑器打开 3. **菜单打开**——编辑器内按 `Ctrl+O` 或点击"打开图片"按钮选择本地文件 4. **拖拽与粘贴**——空窗口状态下，把图片文件拖入窗口，或按 `Ctrl+V` 粘贴剪贴板中的图片即可开始编辑 窗口会根据图片的实际尺寸自动调整大小，避免小图撑满屏幕或大图显示不全。 ## 标注工具 工具栏位于窗口顶部，每个工具都有独立快捷键（窄窗口下部分工具收纳进"更多工具"下拉菜单）： | 快捷键 | 工具 | 说明 | | --- | --- | --- | | `V` | 选择 | 选中、移动、缩放已有标注对象；按住 `Space` 拖拽可平移画布 | | `1` | 矩形 | 框选重点区域，可调圆角与填充 | | `2` | 椭圆 | 圆形/椭圆标注 | | `3` | 箭头 | 指向性标注，箭头大小可调 | | `4` | 直线 | 直线标注 | | `5` | 画笔 | 自由手绘 | | `6` | 高亮 | 半透明荧光笔，突出文字 | | `7` | 马赛克 | 涂抹打码，块大小可调 | | `8` | 文字 | 添加文字说明，字号可调 | 绘制后对象仍然可以编辑：用选择工具点中对象，即可移动、缩放，或在属性面板中修改它的颜色、线宽等参数。 ## 图像处理 工具栏在标注工具右侧提供一组图像处理动作，直接作用于底图像素、同步变换已有标注，并全部接入撤销历史： - **裁剪**（`C`）——激活后画布上出现裁剪框，拖拽 8 个手柄或整体移动来选择保留区域；浮动操作条可选比例预设（自由 / 1:1 / 4:3 / 3:4 / 16:9 / 9:16），或点击"自动裁边"一键检测内容包围盒、快速裁掉四周的纯色白边。应用后裁掉多余部分，按 `Esc` 取消 - **旋转**——顺时针 / 逆时针 90° 旋转整张图片，标注对象同步跟随 - **翻转**——水平 / 垂直镜像翻转 - **缩放**——对话框输入目标宽高，可锁定比例联动，另有 50% / 75% / 100% / 150% / 200% 百分比快捷档，底图高质量重采样 - **背景智能透明**——点击即自动识别图片四边的背景色（白、黑或任意纯色），只把与边界连通的背景抠成透明，图内的白色文字、色块不受影响；并对边缘做抗锯齿羽化，消除圆角、斜边残留的背景色晕边，导出 PNG 边缘干净 自动裁边与背景透明共用**背景容差**：在右侧属性面板未选中任何对象时的"图像处理"区拖动滑杆（0–128，默认 16）调节判定灵敏度——背景色不纯或有渐变时调大，需要更精确保留时调小。透明区域在画布上以棋盘格显示，导出为真透明。 ## 属性与图层 右侧属性面板（窄窗口下为抽屉式）分两部分： - **样式属性**——颜色、自定义颜色、填充色、不透明度、描边宽度；文字工具可调节字号，矩形可调节圆角，马赛克可调节块大小，箭头可调节箭头大小。选中已有对象时，修改会直接作用于该对象 - **对象操作**——复制、删除、置顶、置底 - **图像处理**——未选中任何对象时显示背景容差滑杆（0–128），供自动裁边与背景智能透明共用 面板底部的**图层列表**展示图片上的全部标注对象（按叠加顺序排列），支持： - 切换单个图层的可见性（临时隐藏某个标注） - 上移 / 下移调整叠加顺序 ## 查看与导航 - **缩放**——底部栏提供放大 / 缩小按钮，另有"适应屏幕"（`Ctrl+0`）与"实际大小"（`Ctrl+1`）两种预设视图 - **平移**——选择工具下按住 `Space` 拖拽，快速查看大图细节 - **撤销 / 重做**——`Ctrl+Z` / `Ctrl+Y`，覆盖绘制、移动、删除、属性修改等全部操作 - 底部栏实时显示当前缩放比例与图片尺寸 ## 保存与分享 - **复制到剪贴板**（`Ctrl+Shift+C`）——把编辑结果作为图片复制，粘贴到任何支持图片的位置，按钮会短暂显示"已复制"确认 - **保存**（`Ctrl+S`）——覆盖保存到原文件路径；截图等尚无文件的图片会自动转为"另存为" - **另存为**（`Ctrl+Shift+S`）——选择新路径与文件名保存副本 通过 Launcher 打开的图片在编辑器内"原地编辑"：保存直接覆盖原文件，不产生副本，符合"快速修一下"的直觉。 ## 图片信息与 EXIF 底部栏的"信息"按钮打开图片信息面板，展示两类数据： - **基本信息**——格式、文件大小、分辨率、色彩类型、位深、每像素位数、Alpha 通道、DPI - **EXIF 信息**（照片自带时）——制造商、型号、镜头、拍摄时间、ISO、曝光时间、光圈、焦距与等效焦距、曝光程序、测光模式、闪光灯、白平衡、色彩空间、GPS 经纬度与海拔、软件、作者、版权等 排查"这张照片是什么时候、用什么设备拍的"这类问题时非常实用。 ## 快捷键一览 | 快捷键 | 功能 | | --- | --- | | `V` / `1` ~ `8` | 切换选择工具与 8 种标注工具 | | `C` | 裁剪工具 | | `Space` + 拖拽 | 平移画布 | | `Ctrl+Z` / `Ctrl+Y` | 撤销 / 重做 | | `Ctrl+S` / `Ctrl+Shift+S` | 保存 / 另存为 | | `Ctrl+O` | 打开图片 | | `Ctrl+Shift+C` | 复制到剪贴板 | | `Ctrl+0` / `Ctrl+1` | 适应屏幕 / 实际大小 | > **💡 macOS 用户** > 上表中的 `Ctrl` 在 macOS 上对应 `Cmd`。 ## 延伸阅读 - [截图与录屏](https://yeero.ai/docs/zh/capture.html)——图片编辑器最常见的入口 - [Launcher](https://yeero.ai/docs/zh/launcher.html)——用路径或粘贴快速打开图片 - [视频编辑器](https://yeero.ai/docs/zh/video-editor.html)——需要处理视频时的姊妹工具 --- # 视频编辑器 *Yeero Desktop 内置的桌面视频剪辑工具：录屏结束直接进入，也可以打开任意本地视频。多轨时间轴、7 种转场、滤镜调色、文字字幕、音频处理与 AI 智能剪辑，录完即剪，剪完即发。* > **⚡ 一句话摘要** > 内置视频编辑器覆盖录完即剪、剪完即发：多轨时间轴、7 种转场、滤镜调色、文字字幕、音频处理与 AI 智能剪辑。 ## 概览 视频编辑器以独立窗口运行，界面分为三个区域： - **预览区**——实时播放画面，支持全屏、适应屏幕与 A-B 循环播放 - **时间轴**——视频轨、标注轨、音频轨、音乐轨多轨并列，带缩略图与波形，支持滚轮缩放与高度调整 - **属性面板**——选中片段或标注后，在此调整全部参数 关于视频格式的几点说明： - 编辑流程基于 **H.264 编码的 MP4**；打开其他编码的视频时，编辑器会自动转码为 H.264 后即可正常播放和编辑 - 首次转码或导出时，会自动下载视频处理工具（ffmpeg），只需等待一次 - 关闭窗口或打开新视频时，若存在未导出的修改会弹出确认对话框，避免误丢工作成果 ## 打开视频 - **录屏后进入**——[录屏](https://yeero.ai/docs/zh/capture.html)结束后点击"用视频编辑器打开"，录好的视频直接载入 - **菜单打开**——点击"打开视频"按钮选择本地文件，也可以选择"用新窗口打开"并行处理多个视频 - **拖拽**——把视频文件直接拖入编辑器窗口 - **Launcher**——在 Launcher 中输入或粘贴视频文件路径，回车即打开 ## 时间轴剪辑 时间轴是编辑的核心。每个视频片段带缩略图，可以： - **分割**——按 `S` 在播放头位置把片段一分为二 - **删除 / 波纹删除**——`Delete` 删除后留空隙；`Shift+Delete` 波纹删除，后续片段自动前移补位 - **修剪**——拖动片段左右边缘的裁剪手柄，去掉片头片尾（裁剪量精确到左右各一段） - **变速**——选中片段在属性面板中调整播放速度，支持自定义倍速，片段的时长会随速度同步变化 - **入点 / 出点**——按 `I` / `O` 标记选区，`X` 清除，`Ctrl+Delete` 删除选区范围 - **复制片段**——复制后粘贴出重复段落 视图方面：`Ctrl+↑` / `Ctrl+↓` 缩放时间轴，`Ctrl+0` 一键适应整个时间轴，滚轮也可缩放；拖动片段时会自动吸附对齐到相邻片段边缘。 ## 转场效果 在相邻片段之间添加转场，让衔接更自然。内置 7 种转场： | 转场 | 效果 | | --- | --- | | 淡入淡出 | 经典的交叉淡化 | | 左滑 / 右滑 / 上滑 | 新画面从对应方向滑入 | | 缩放 | 画面缩放过渡 | | 闪白 | 白场闪烁过渡，适合节奏切换 | | 模糊 | 模糊化过渡 | 转场时长可在 **0.1 ~ 3.0 秒**之间调节，默认 0.5 秒。 ## 滤镜与调色 - **滤镜**——胶片、复古、清新、电影、暖色、冷色、黑白共 7 种风格滤镜，另有"原始"一键还原 - **调色**——亮度、对比度、饱和度、色温、色调、曝光、高光、阴影八项参数精细调节，支持一键重置调色 ## 文字与字幕 - **添加文字**——在叠加轨道放置文字层，可调字体、字号、字重、文字颜色、透明度、描边与阴影，水平 / 垂直对齐自由设定 - **文字模板**——内置 Vlog 标题等模板，一键套用版式 - **文字动画**——打字机、淡入、弹跳、上滑四种出场动画 - **字幕**——支持导入字幕文件与自动字幕 ## 视频标注 与 [图片编辑器](https://yeero.ai/docs/zh/image-editor.html)同源的 9 种标注工具（选择、矩形、椭圆、箭头、直线、画笔、高亮、马赛克、文字），区别在于每个标注都带时间范围：设定入点与出点后，标注只在指定时间段内显示，适合"第 3 秒圈出问题按钮"这类讲解场景。标注存放在独立的标注轨道，可在时间轴上直接调整出现时机。 ## 音频处理 - **原声音频**——视频自带音频显示在音频轨（带波形），可分割、复制、删除 - **音量控制**——单个音频片段音量、静音 / 取消静音，另有全局主音量 - **淡入淡出**——为音频片段添加淡入、淡出效果 - **分离音频**——把视频音轨分离为独立音频片段单独处理 - **背景音乐**——独立的音乐轨道，可添加本地音频文件作为 BGM ## 智能剪辑 不想逐帧剪？智能剪辑自动分析全片，找出冗余片段并给出处理建议： - **静止**——画面长时间无变化的段落 - **静音**——长时间无声的段落 - **重复**——内容重复的段落 检测结果列出每一段的类型与时长，你可以为每段选择动作：**剪掉**、**加速**或**忽略**，也可以全部确认 / 全部忽略。面板实时显示"可节省时长"，点击任意段落可跳转到该位置预览。确认无误后点击"应用裁剪"，冗余段落一次性从时间轴移除。 ## 导出 按 `Ctrl+E` 或点击"导出"打开导出对话框，可配置： - **导出预设**——预设组合一键选定 - **分辨率**——可选原始分辨率或指定分辨率 - **帧率与格式**——按需调整 - **输出路径**——自由选择保存位置 导出过程显示实时进度，完成后提示并可打开所在文件夹。另外，`Ctrl+Shift+F` 可随时截取当前帧复制到剪贴板，方便在聊天中分享画面。 ## 快捷键一览 编辑器内按 `Shift+?` 可随时呼出快捷键面板。 | 快捷键 | 功能 | | --- | --- | | `Space` | 播放 / 暂停 | | `←` / `→` | 上一帧 / 下一帧（暂停时） | | `Shift+←` / `Shift+→` | 快退 / 快进 5 秒 | | `S` | 在播放头处分割片段 | | `Delete` / `Shift+Delete` | 删除 / 波纹删除 | | `I` / `O` / `X` | 设入点 / 设出点 / 清除入出点 | | `Ctrl+Delete` | 删除入出点之间的范围 | | `Ctrl+A` / `Escape` | 全选 / 取消选择 | | `Ctrl+Z` / `Ctrl+Shift+Z` | 撤销 / 重做 | | `Ctrl+↑` / `Ctrl+↓` / `Ctrl+0` | 时间轴放大 / 缩小 / 适应 | | `V` | 切换到选择工具 | | `Ctrl+E` | 导出视频 | | `Ctrl+Shift+S` | 智能剪辑 | | `Ctrl+Shift+F` | 截取当前帧到剪贴板 | > **💡 macOS 用户** > 上表中的 `Ctrl` 在 macOS 上对应 `Cmd`。 ## 延伸阅读 - [截图与录屏](https://yeero.ai/docs/zh/capture.html)——视频素材的第一入口 - [图片编辑器](https://yeero.ai/docs/zh/image-editor.html)——同源的标注工具体系 - [Launcher](https://yeero.ai/docs/zh/launcher.html)——快速打开视频文件或唤起编辑器 --- # 磁盘清理 *用目录树可视化磁盘空间占用，自动识别 30+ 分类的应用，AI 分析未知目录，安全清理到回收站——全部在 Yeero Desktop 中完成。* > **⚡ 一句话摘要** > 磁盘清理工具扫描磁盘、用树形视图展示每个文件夹的大小、自动识别应用，让你安全地释放空间——文件先移到回收站，30 天内可恢复。 > **💡 仅桌面端** > 磁盘清理需要访问本地文件系统，依赖 Tauri 的原生能力，因此仅在 Yeero Desktop 中可用，网页端不支持。 ## 概述 磁盘清理帮你回答一个问题："到底是什么在吃我的磁盘空间？"它将可视化目录树与内置应用识别、可选的 AI 分析结合，让你从"这个文件夹好大"到"我知道它是什么、能不能删"只需几秒钟。 - **多磁盘支持**——扫描本地磁盘、U 盘、网络驱动器和 RAM 磁盘，一键切换扫描目标 - **目录树**——逐层展开文件夹，每个节点显示其总大小 - **应用识别**——已识别的应用会标注名称和分类（30+ 内置分类） - **AI 分析**——未知目录可交给 AI 分类，返回应用名、分类、数据类型和置信度 - **安全清理**——文件移到回收站而非永久删除，30 天内可随时恢复 ## 启动方式 有两种方式打开磁盘清理： 1. **通过 Launcher**——按 `Alt+Space`，输入"磁盘清理"或"disk cleaner"，回车即可 2. **通过主窗口**——在桌面端功能区找到磁盘清理入口 首次启动时会弹出一个简短的引导对话框，介绍三个核心步骤：浏览、识别、清理。你可以关闭并选择不再显示。 ## 磁盘概览 顶部栏为每个检测到的磁盘显示一张卡片： - **磁盘名称和盘符**——如"本地磁盘 (C:)" - **使用量条**——水平进度条按填充比例着色，超过 85% 变红 - **已用 / 总量**——如"186.3 GB / 500 GB"，附带百分比 - **驱动器类型图标**——本地、USB、网络、光盘和 RAM 磁盘各有不同图标 点击任意卡片即可切换扫描目标到该磁盘。刷新按钮重新加载概览。使用率超过 90% 的磁盘还会在概览栏下方触发警告横幅。 ## 目录树浏览 选中磁盘后，清理工具扫描其顶级目录并以树形视图展示： - **点击展开**——每个文件夹展开子目录及其大小，逐层定位空间大户 - **大小条**——每个节点显示比例条，表示它占父目录的比例 - **扫描进度**——顶部进度条显示当前路径、已扫描文件数和总大小，可随时取消 - **缓存与刷新**——扫描结果会缓存以便快速重新进入；时间戳显示数据捕获时间，"重新扫描"按钮强制刷新。后台刷新时树形视图保持可见 ## 应用识别 清理工具内置了一个数据库，可识别 30+ 分类的常见应用： | 分类组 | 示例 | | --- | --- | | 浏览器 | Chrome、Firefox、Edge、Safari、Arc | | 开发工具 | VS Code、IntelliJ、Xcode、Android Studio、Docker | | 设计 | Photoshop、Figma、Sketch、Blender | | 媒体 | Spotify、VLC、OBS、Premiere Pro | | 通讯 | 微信、Slack、Discord、Teams | | 系统与工具 | Homebrew、npm、pip、conda | 已识别的文件夹会标注应用标签，显示名称和分类。每个条目还带有数据类型标签（缓存、日志、用户数据、配置、安装包、临时文件、备份、媒体缓存、包缓存等）和清理建议级别。 ### AI 分析 对于内置数据库无法识别的文件夹，可以点击"AI 分析"让模型识别： - **单个分析**——一次分析一个目录 - **批量分析**——选中多个未知目录一起分析（消耗积分） - **结果详情**——AI 返回应用名、分类、数据类型和置信度（高 / 中 / 低） - **反馈机制**——标记每个结果为正确或错误，帮助提升后续准确度 AI 结果与内置识别结果并排显示，视图保持一致。 ## 清理操作 确定要清理的内容后： 1. **选择项目**——勾选要清理的文件夹 2. **查看摘要**——底部面板显示已选项目数和预计释放空间 3. **确认清理**——点击"安全清理"并在确认对话框中确认 根据清理的大小和类型，可能会出现额外的确认步骤： - **大文件警告**——清理大量数据时需要明确确认 - **运行中应用警告**——如果相关应用仍在运行，会要求先关闭 - **聊天文件警告**——聊天相关文件需要输入确认词，因为清理后无法恢复 ## 清理安全策略 每个可清理项目都标记了三个安全级别之一： | 级别 | 含义 | | --- | --- | | **安全清理** | 缓存、临时文件等可以无副作用删除的数据 | | **谨慎清理** | 可能需要重新下载的数据（如包缓存、媒体缓存） | | **不建议清理** | 用户数据或系统关键文件——清理可能导致数据丢失或应用异常 | > **⚠ 回收站保护** > 所有清理的文件都会移到系统回收站，而非永久删除。你有 30 天的时间恢复任何改变主意的文件。 清理完成后，结果对话框会显示：成功清理的项目数、释放的空间，以及失败的项目（附带原因，如路径不存在或系统保护路径）。 ## 清理历史 清理历史面板记录每次清理： - **每次详情**——日期、清理项目数和释放空间 - **累计统计**——总释放空间和总清理次数，显示在磁盘概览下方的统计栏中 - **删除记录**——可按需删除单条历史记录 ## 延伸阅读 - [Launcher](https://yeero.ai/docs/zh/launcher.html)——通过 Alt+Space 一键打开磁盘清理 - [桌面端快速入门](https://yeero.ai/docs/zh/desktop.html)——Yeero Desktop 入门指南 - [文件](https://yeero.ai/docs/zh/files.html)——管理云端文件和自动清理规则 --- # 文件 *完整的文件管理系统，支持上传、预览、组织、语义搜索和自动清理规则。* > **⚡ 一句话摘要** > 文件库统一管理你的上传内容：在线预览、文件夹组织、AI 语义搜索，以及自动清理规则。 ## 文件概述 文件页面为你上传的和 AI 生成的所有内容提供全面的文件管理系统。 ## 上传文件 要上传文件： 1. 点击**"上传文件"**或使用拖放。 2. 从电脑中选择文件。支持任何格式，单个文件最大 **512 MB**。 3. 选择目标文件夹（可选）。 4. 点击**"上传"**开始上传。 > **💡 拖放上传** > 你也可以直接将文件拖放到文件页面上立即开始上传。 ## 文件预览 点击任何文件上的**预览**按钮打开通用文件查看器： - **图片** — 查看，支持放大/缩小和重置控件。 - **视频** — 播放，标准视频控件。 - **音频** — 播放，静音/取消静音控件。 - **文档** — 查看支持的文档格式。 - **下载** — 下载任何文件到你的电脑。 如果文件类型不支持预览，系统会提示你下载。 ## 组织文件 ### 查看模式 - **网格视图** — 以缩略图网格形式显示文件。 - **列表视图** — 以详细列表格式显示文件。 ### 筛选 - **来源** — 全部、我的上传或 AI 创建。 - **类型** — 所有类型、图片、视频、音频、文档、其他。 - **排序** — 最新优先或最早优先。 ### 文件操作 - **重命名** — 更改文件名。 - **移动** — 使用可视化文件夹浏览器移动到其他文件夹。 - **删除** — 永久删除文件。 - **前往对话** — 导航到创建文件的对话。 ## 语义搜索 点击搜索栏旁边的切换图标，从文件名搜索切换到 **AI 语义搜索**。用自然语言描述你要找的文件，AI 会基于内容相似度找到匹配的文件。 ## 存储管理 YeeroAI 提供 **1 GB 免费存储**。如果超出免费额度： - 超额存储会产生每日积分费用。 - 如果余额不足，你会看到警告横幅。 - 如果逾期超过 7 天，最旧的文件将被自动删除，直到存储降到 1 GB。 > **⚠ 自动清理** > 如果存储逾期 7 天以上，系统将自动删除最旧的文件，直到你回到免费额度内。请充值积分或手动删除文件以避免此情况。 ## 清理规则 使用清理规则自动删除文件。规则每天凌晨 2 点运行。 ### 创建规则 1. 点击文件页面上的**清理**按钮。 2. 点击**"新建规则"**。 3. 配置规则：**规则名称** — 描述性名称（如"删除旧的 AI 生成图片"）。 4. **媒体类型** — 按图片、视频、音频等筛选。 5. **文件来源** — 按上传或 AI 创建筛选。 6. **上传超过** — 最少天数（7-3650天）。 7. **文件大小 ≥** — 最小文件大小（MB）。 8. **路径范围** — 使用通配符定位特定文件夹路径（如 `/photos/**`）。 ### 管理规则 - **启用/禁用** — 切换规则开关。 - **编辑** — 修改规则条件。 - **删除** — 永久删除规则。 - **预览** — 重新运行预览查看当前匹配的文件。 - **立即删除** — 手动立即执行规则。 ## 批量操作 文件库列表支持鼠标拖拽矩形框选，快速圈选多个文件后进行批量操作。 在列表空白区域按住鼠标左键拖拽生成框选框，松开后范围内文件自动高亮，页面顶部浮层展示可用操作：**批量下载**、**批量删除**、**批量移动到文件夹**。按 `Esc` 退出多选模式。 批量下载同时支持文件和文件夹。在浏览器中，文件夹内的文件会摊平下载；在 Tauri 桌面端，下载时会保持原目录结构。 --- # 文件夹 *嵌套文件夹层级，支持置顶、排序、移动操作和对话、应用、消息的参数继承。* > **⚡ 一句话摘要** > 文件夹以嵌套层级组织对话、应用、提示词与知识：支持置顶、排序、批量移动，以及文件夹级参数继承。 ## 文件夹概述 YeeroAI 中的文件夹帮助你以嵌套层级组织对话、应用、消息、记忆和提示词。文件夹系统支持： - **三种文件夹类型** — 对话文件夹、应用文件夹和消息文件夹。 - **嵌套结构** — 在文件夹内创建子文件夹进行深层组织。 - **参数继承** — 在文件夹级别设置模型、工具、记忆和生成参数，子项目会继承它们。 ## 导航文件夹 ### 面包屑导航 顶部的面包屑显示你当前的文件夹路径。点击任何片段可以向上导航。使用面包屑中的文件夹搜索快速找到子文件夹。 ### 侧边栏导航 侧边栏的文件夹页面提供对所有三种文件夹类型的访问，带有搜索、排序和筛选选项。 ## 文件夹操作 | 操作 | 描述 | | --- | --- | | 新建文件夹 | 创建一个带有名称和可选描述的新文件夹。 | | 编辑 | 修改文件夹名称和描述。 | | 置顶 / 取消置顶 | 将文件夹置顶到列表顶部以便快速访问。 | | 移动 | 将文件夹移动到另一个文件夹中。 | | 删除 | 永久删除文件夹及其所有内容。 | | 启用/禁用记忆 | 切换文件夹中所有项目的 AI 记忆。 | | 启用/禁用提示词 | 切换文件夹的提示词模板。 | ## 参数继承 文件夹支持参数继承，允许你设置所有子项目（对话、子文件夹）继承的默认值： - **模型** — 此文件夹中对话的默认模型选择。 - **工具** — 默认工具选择。 - **记忆** — 为文件夹启用或禁用 AI 记忆。 - **上下文摘要** — 使用摘要而非完整内容。 - **生成参数** — 温度、采样、上下文窗口等。 ### 继承模式 - **继承** — 使用父文件夹的设置（默认）。标有"继承"标记。 - **自定义** — 使用此文件夹的特定自定义设置。点击"切换到自定义"进行覆盖。 - **重置为继承** — 返回从父文件夹继承。 ## 文件夹中的内容 文件夹可以包含以下类型的内容： | 类型 | 描述 | | --- | --- | | 文件夹 | 嵌套子文件夹。 | | 对话 | 一个包含分支和消息的对话。 | | 消息 | 单条消息（用户或 AI）。 | | 记忆 | AI 生成或手动创建的记忆条目。 | | 提示词 | 可复用的提示词模板。 | | 应用 | AI 驱动的应用程序。 | ## 文件夹限制 | 限制 | 值 | | --- | --- | | 最大文件夹数 | 1,000 | | 文件夹名称唯一性 | 同一父文件夹内名称必须唯一。 | | 不能移动到后代 | 文件夹不能移动到自己的子文件夹中。 | ## 批量操作 文件夹内的引用列表支持鼠标拖拽矩形框选，快速圈选多个内容后进行批量操作。 在列表空白区域按住鼠标左键拖拽生成框选框，松开后范围内内容自动高亮，页面顶部浮层展示可用操作：**批量删除**、**批量移动**。按 `Esc` 退出多选模式。 --- # Prompt 模板 *创建和管理可复用的提示词模板，支持启用/禁用切换。一键将任何提示词插入对话中。* > **⚡ 一句话摘要** > 提示词模板把你最好的提示词存下来，一键插入任意对话：支持启用/禁用切换与文件夹分类管理。 ## 什么是 Prompt 模板？ Prompt 模板是预先编写的提示词，你可以快速插入到任何对话中。它们帮助你： - **节省时间** — 复用常用提示词而无需重新输入。 - **确保一致性** — 在对话中使用标准化的提示词。 - **分享最佳实践** — 为常见任务和工作流创建提示词。 ## 创建提示词 1. 导航到**设置 → 提示词**。 2. 点击**"新建提示词"**。 3. 填写表单：**提示词名称** — 描述性名称。 4. **提示词内容** — 提示词文本，支持 Markdown。 5. **文件夹** — 选择保存提示词的文件夹。 ## 管理提示词 ### 启用 / 禁用 切换提示词的开关。禁用的提示词不会出现在快速插入菜单中，但不会被删除。 ### 搜索和筛选 - **搜索** — 按名称搜索提示词。 - **状态筛选** — 显示全部、已启用或已禁用。 - **排序** — 最近更新或最早创建。 ### 编辑和移动 - **编辑** — 修改提示词名称、内容或文件夹。 - **移动** — 将提示词移动到其他文件夹。 - **删除** — 永久删除提示词。 ### 预览 点击**"预览"**查看提示词内容如何以 Markdown 格式渲染。 ## 批量操作 ### 批量移动 1. 点击**"选择"**进入选择模式。 2. 选择你要移动的提示词。 3. 点击**"批量移动"**并选择目标文件夹。 ### 批量删除 你可以通过两种方式批量删除提示词： - **删除所有已禁用** — 一次性移除所有已禁用的提示词。 - **选择并删除** — 选择特定提示词并批量删除。 ## 在对话中使用提示词 在对话中撰写消息时，你可以从库中插入已启用的提示词。提示词内容将添加到你的消息输入中，随时可以发送。 ## 预设角色提示词库 除了自己创建提示词，YeeroAI 还内置了一套**官方预设角色提示词库**，覆盖 25 大类、242 个精心编写的角色（中英双语），帮你快速获得专业级提示词，无需从零撰写。 ### 浏览与引入 1. 进入**设置 → 提示词**，切换到**“预设库”**标签页。 2. 按分类筛选，或用关键词搜索（支持名称、描述与别名）。 3. 点击**“预览”**查看角色提示词全文。 4. 点击**“使用此预设”**，选择保存的文件夹，即可一键引入到**“我的提示词”**。 ### 官方更新与同步 引入的预设会记录其来源与官方版本。当官方发布了更新版本时，对应的提示词卡片上会出现**“官方有更新”**标记；点击**“同步官方最新”**即可将其内容更新到最新官方版。 > **⚠️ 同步会覆盖本地修改** > 同步会用官方最新版本**覆盖**你当前的名称与内容，且不可撤销。如已自行修改，请先备份。 引入后的预设就是普通的**我的提示词**，可以自由编辑、移动、启用/禁用与删除，编辑不会影响官方原件。 ## AI 辅助创作 在新建或编辑提示词时，你可以借助 **AI 辅助**快速生成或打磨内容，而不必从空白开始。 - **AI 生成** — 描述你想要的角色或用途，由 AI 直接生成一份结构化提示词。 - **改写 / 润色 / 扩写 / 精简** — 对已有内容进行改写、润色、扩写或精简。 - **模型选择** — 可选择用于辅助的模型，或使用自动选择。 AI 辅助的结果会先展示在预览区，确认满意后再点击**“应用”**写入提示词内容，你可以继续手动修改。 > **ℹ️ 关于消耗** > AI 辅助会按次消耗积分，具体以界面提示为准。 ## 提示词限制 | 限制 | 值 | | --- | --- | | 最大提示词数 | 1,000 | | 名称必须唯一 | 提示词名称不能重复。 | --- # 设置 *管理主题、语言、默认模型、记忆、后台流式传输、云同步和所有应用偏好设置。* > **⚡ 一句话摘要** > 设置页集中管理 YeeroAI 的一切行为偏好：主题、语言、默认模型、记忆、上下文长度、后台流式传输与云同步。 ## 设置概述 从侧边栏访问设置。设置按以下部分组织： - **系统** — 主题、语言、端到端加密密钥等基本配置。 - **通用** — 自动对话命名、上下文窗口、记忆、默认模型、推理力度。 - **后台流式传输** — 导航离开时继续 AI 生成。 - **云同步** — 跨设备同步设置和应用数据。 - **提示词** — 管理提示词模板。 - **记忆** — 管理 AI 记忆条目。 ## 系统设置 ### 主题 在**浅色**和**深色**主题之间切换。你的偏好会被保存并在会话间持久化。 ### 语言 在**English**和**中文**之间切换界面语言。界面会立即更新。 ### 端到端加密密钥 配置通用端到端加密密钥，服务于所有加密同步场景（剪贴板跨设备同步、Agent 消息等）。密钥仅保存在你本机的系统密钥链中，留空表示不开启加密。 - **一键生成** — 不想自己记密钥时，可让应用自动生成一个高强度密钥。 - **多设备保持一致** — 要在另一台设备上解密已同步的加密内容（如剪贴板条目），请在每台设备上配置相同的密钥。 ## 通用设置 | 设置 | 描述 | | --- | --- | | 自动生成对话名称 | 根据对话内容自动生成对话标题。 | | 上下文窗口大小 | 对话中保留的历史消息数量。必须是有效范围内的数字。 | | 启用记忆 | 允许 AI 记住对话中的关键信息。可在对话或文件夹级别覆盖。 | | 默认模型 | 为所有新对话设置默认模型。最多选择 10 个模型。 | | 推理力度 | 默认推理力度级别：无、最低、低、中、高或极高。 | | 上下文摘要 | 构建上下文时使用消息摘要而非完整内容。减少 token 使用量但可能丢失一些细节。 | ## 后台流式传输 当**后台流式传输**启用时： - 即使你导航离开对话，AI 生成也会继续。 - 页面刷新或网络中断后响应会被保留。 - 当你返回对话时，你会收到任何剩余内容。 > **💡 弹性恢复** > 后台流式传输能优雅地处理网络中断。即使你的连接暂时中断，你也不会丢失任何已生成的内容。 ## 云同步 ### 同步项目设置 将你的主题和其他本地偏好同步到云端，使其在所有设备上保持一致。 ### 同步应用数据 将 AI 应用运行时存储数据同步到云端。你可以查看哪些应用已禁用同步并单独重新启用它们。 ## 记忆管理 设置中的**记忆**部分让你管理所有 AI 记忆条目： - **查看所有记忆** — 包括自动生成和手动创建的。 - **搜索** — 按名称或内容搜索记忆。 - **按状态筛选** — 已启用、已禁用、嵌入中或已遗忘。 - **按特质筛选** — 固定、非常稳定、用户习惯、用户偏好、可变或弱。 - **启用 / 禁用** — 切换单个记忆。 - **编辑** — 修改记忆名称和内容。 - **移动** — 移动到其他文件夹。注意：来自对话的自动生成记忆不能移动。 - **删除** — 永久删除记忆。 - **批量操作** — 批量移动和批量删除记忆。 ## 批量操作 设置中心的提示词列表和记忆列表均支持鼠标拖拽矩形框选，快速圈选多条后进行批量操作。 在列表空白区域按住鼠标左键拖拽生成框选框，松开后范围内内容自动高亮，页面顶部浮层展示可用操作。按 `Esc` 退出多选模式。 | 列表 | 可用批量操作 | | --- | --- | | 提示词列表 | 批量删除、批量移动到文件夹 | | 记忆列表 | 批量删除、批量移动到文件夹 | ## 账户设置 从账户页面，你可以管理： - **个人资料** — 编辑昵称、用户名、简介和上传头像。 - **账户安全** — 绑定/解绑邮箱、Google 和 GitHub 账户。 - **退出登录** — 登出并清除本地登录状态。 ### 登录方式 你可以绑定多种登录方式以提高安全性： - **邮箱** — 通过验证绑定或更改邮箱地址。 - **Google** — 关联或解除关联你的 Google 账户。 - **GitHub** — 关联或解除关联你的 GitHub 账户。 > **⚠ 至少保留一种方式** > 你必须至少保留一种登录方式绑定到你的账户。系统会阻止你解绑最后一种方式。 --- # 计费与积分 *简单透明的按需计费——注册即送积分，每次消耗实时可见，随时充值，随时取消。* > **⚡ 一句话摘要** > YeeroAI 全程用积分按量计费：注册送 1,000 积分（30 天有效），每条回复的消耗实时可见，一次性购买的积分永不过期，购买后 24 小时内可按剩余比例退款。 ## 积分体系总览 YeeroAI 的所有 AI 能力都用同一种**积分（Credits）**结算：没有强制月费，用多少扣多少。 - **注册即送 1,000 积分** — 30 天有效，无需绑定信用卡。 - **消耗实时可见** — 每条 AI 回复都标注本次消耗的积分数，不会在回复之外额外扣费。 - **免费模型零消耗** — 在模型选择器中打开"仅免费"，即可使用不消耗积分的模型。 - **余额随时可查** — 在**账户 → 积分记录**查看余额、收支明细与账单。 - **余额不足时** — AI 请求会失败并提示充值，已有的对话、文件与知识条目不会因此丢失。 > **💡 消耗多少由什么决定** > 单次消耗取决于所用模型与输出长度。同一条消息选择多个模型并行回答时，每个模型分别计费。 ## 积分有效期 不同来源的积分有效期不同，过期后不可恢复、也不可退款： | 积分来源 | 有效期 | 说明 | | --- | --- | --- | | 注册赠送 | 30 天 | 注册即送的 1,000 积分，到期未用完的部分以"积分过期"记录扣除。 | | 充值购买（一次性） | 永不过期 | 一次性购买的积分长期有效，可按需囤积。 | | 订阅档位每月积分 | 当期 30 天 | 随订阅周期发放，订阅到期或取消后当期积分失效。 | | 邀请奖励 | 永不过期 | 通过邀请好友获得的积分长期有效。 | | 活动奖励 | 以活动说明为准 | 促销与活动赠送的积分，有效期在积分记录中标注。 | 每笔积分的具体到期时间，可在**账户 → 积分记录 → 获取记录**的"过期时间"列查看，长期有效的会直接标注**"永不过期"**。 ## 获取积分 | 方式 | 说明 | | --- | --- | | **注册奖励** | 注册即送 1,000 积分（30 天有效）。 | | **充值购买** | 按需购买积分包或订阅档位，详见定价与套餐。 | | **邀请奖励** | 好友注册及其每次购买积分，你都能获得 10% 奖励，无上限且永不过期，详见邀请奖励。 | | **活动奖励** | 来自特别促销与官方活动的积分。 | ### 获取记录怎么看 在**账户 → 积分记录 → 获取记录**中，每笔入账都完整留痕： | 字段 | 含义 | | --- | --- | | 获取方式 | 注册奖励、充值购买、邀请奖励或活动奖励。 | | 时间 | 积分入账时间。 | | +积分 | 本笔入账的积分数量。 | | 获取前余额 | 入账前的积分余额，便于逐笔核对。 | | 剩余额度 | 这笔积分当前还剩多少未使用。 | | 过期时间 | 这笔积分的到期时间，或标注"永不过期"。 | | 备注 | 来源补充说明，如活动名称或订单信息。 | | 操作 | 充值购买的记录可在此申请退款。 | 支持按日期范围筛选（最多回溯一年），列表滚动到底部会自动加载下一页。 ## 消耗积分 ### AI 调用 | 方式 | 说明 | | --- | --- | | **对话消耗** | 每条 AI 回复按模型与输出长度计费，消耗数直接显示在回复上。 | | **应用生成** | AI 应用生成代码与内容时消耗。 | | **应用讨论** | 在 AI 应用内进行讨论协作时消耗。 | | **Agent 模型调用** | Agent 模式下自动调用模型产生的消耗。 | | **Cloud Agent 容器运行** | 云端 Agent 任务运行所占用的容器资源费用。 | | **磁盘清理分析** | 桌面端磁盘清理的 AI 分析，执行前会明确提示"消耗 N 积分"。 | ### 记忆与知识 | 方式 | 说明 | | --- | --- | | **生成记忆** | AI 从对话中提炼知识条目时消耗。 | | **匹配记忆** | 为新消息匹配相关记忆上下文时消耗。 | ### 存储 | 方式 | 说明 | | --- | --- | | **文件存储费用** | 文件库占用超出 **1 GB** 免费额度后，对超额部分按 **50 积分 / GB / 天**计费，详见存储计费。 | | **云端文档存储费用** | AI 应用云端文档占用超出 **10 MB** 免费额度后，对超额部分按 **5 积分 / 10 MB / 天**计费，详见存储计费。 | ### 其他 | 方式 | 说明 | | --- | --- | | **积分过期** | 超过有效期未使用的积分被扣除。 | | **退还积分** | 退款成功后，该笔购买对应的剩余积分被收回。 | ### 消耗记录怎么看 在**账户 → 积分记录 → 消耗记录**中，每笔扣费都显示消耗方式、时间、-积分、消耗前余额、消耗后余额与备注，方便逐笔核对某次对话或某个应用的实际开销。同样支持日期范围筛选与滚动加载。 ## 存储计费 YeeroAI 有**两套相互独立的存储额度**：文件库与 AI 应用云端文档。两者额度不互通、分别计费，超出部分都按天从积分余额中扣除。 ### 额度与费率 | 存储类型 | 免费额度 | 超额费率 | 消耗方式 | 用量在哪看 | | --- | --- | --- | --- | --- | | **文件库** | 1 GB | 50 积分 / GB / 天 | 文件存储费用 | 文件页顶部的存储状态条 | | **AI 应用云端文档** | 10 MB | 5 积分 / 10 MB / 天 | 云端文档存储费用 | AI 应用页顶部的存储状态条 | 上表计费的是**账户级的存储总量**。注意与另一项限制区分：AI 应用还有一个"每个应用 1 MB"的云同步运行时数据上限，超出后只会暂停该应用的云同步、需要清除应用数据才能恢复，**不产生任何积分费用**，详见[云同步](https://yeero.ai/docs/zh/ai-apps.html#cloud-sync)。 ### 每天扣多少 只对**超出免费额度的部分**计费，按比例折算后**向上取整**，最低 1 积分： **每日扣费积分 = 超额容量 ÷ 免费额度 × 费率** | 实际占用 | 超额部分 | 计算 | 每日扣费 | | --- | --- | --- | --- | | 文件库 800 MB | 未超额 | — | 0 积分 | | 文件库 1 GB 零 1 MB | 1 MB | 1 MB ÷ 1 GB × 50 ≈ 0.05，向上取整 | 1 积分 | | 文件库 1.5 GB | 0.5 GB | 0.5 × 50 | 25 积分 | | 文件库 3 GB | 2 GB | 2 × 50 | 100 积分 | | 云端文档 8 MB | 未超额 | — | 0 积分 | | 云端文档 15 MB | 5 MB | 5 MB ÷ 10 MB × 5 = 2.5，向上取整 | 3 积分 | | 云端文档 40 MB | 30 MB | 3 × 5 | 15 积分 | ### 扣费规则 - **按天结算** — 超额期间每天从积分余额扣除一次；两套存储各自独立产生一笔消耗记录，可分别核对。 - **额度内零费用** — 删除文件或删除文档、让占用回到免费额度内后，不再产生新的存储费用。 - **页面实时可见** — 状态条直接显示当前每日扣费，文件库形如"25 积分/天"，云端文档形如"当前 +3 积分/天"，无需自己计算。 - **余额可维持天数** — 状态条按"当前余额 ÷ 每日扣费"估算还能支撑几天，不足 3 天时转为红色警告"余额不足 3 天"。 > **💡 云端文档额度比想象中小** > AI 应用云端文档的免费额度只有 **10 MB**，且与文件库的 1 GB 完全独立、不能互相挪用。长期运行、持续产出文档的应用很容易超额，建议在 AI 应用页留意顶部的存储状态条，其悬浮说明会同时给出费率与当前每日扣费。 ### 页面提示对照 | 状态 | 页面提示 | 说明 | | --- | --- | --- | | 额度内 | "存储正常"、"已用 X / 1 GB" | 不产生存储费用。 | | 超出额度 | "已超免费额度"、"超出 X"、"X 积分/天" | 开始按天扣积分，页面实时显示当前每日费用。 | | 余额偏紧 | "余额不足 3 天"、"余额可维持 X 天" | 按当前每日费用估算的可用天数，提醒及时充值。 | | 积分不足 | "存储空间欠费中" | 余额已不足以支付存储费用，进入欠费状态。 | ### 欠费之后会发生什么 欠费状态与 7 天宽限期是**账户级**的，两套存储共用同一套时间线： 1. **开始欠费** — 文件页顶部出现"存储空间欠费中"横幅，并显示已欠费天数。 2. **宽限提醒** — 横幅会告知：若在接下来的若干天内未补充积分，系统将从最早上传的文件开始自动清理，直到存储降至 1 GB。 3. **欠费超过 7 天** — 系统开始自动清理：文件库从最早上传的文件清起，直到回到 1 GB 免费额度内；云端文档同样清理，直到回到 10 MB 免费额度内。 > **⚠ 自动清理不可撤销** > 被自动清理的文件与文档无法恢复。看到欠费横幅后请及时**充值积分**或**手动清理**不需要的内容——两个入口都在横幅上直接提供。 > **⚠ 额度与费率快照说明** > 上表额度与费率为 **2026-09-04** 的快照，可能随运营调整。实际扣费以页面状态条实时显示的每日金额，以及**账户 → 积分记录 → 消耗记录**中的逐笔明细为准。 ## 定价与套餐 套餐分两种形态：**订阅**（按月发放积分，单价更低）与**一次性购买**（买断积分，永不过期）。两者可以并存：订阅覆盖日常用量，一次性购买应对突发需求。 ### 订阅档位 | 档位 | 月费 | 每月积分 | 每积分单价 | | --- | --- | --- | --- | | **Pro** | $5.00 | 37,500 | ≈$0.000133 | | **Pro Plus** | $10.00 | 75,000 | ≈$0.000133 | | **Ultra** | $20.00 | 150,000 | ≈$0.000133 | ### 一次性购买 | 支付渠道 | 单价 | 每份积分 | 每积分单价 | | --- | --- | --- | --- | | Paddle（国际信用卡，美元结算） | $1.00 | 6,000 | ≈$0.000167 | | 微信支付（人民币结算） | ￥1.30 | 1,200 | ≈￥0.00108 | | 支付宝（人民币结算） | ￥1.30 | 1,200 | ≈￥0.00108 | > **💡 订阅比一次性便宜约 20%** > 按每积分单价对比，订阅档约 $0.000133/积分，一次性购买约 $0.000167/积分。用量稳定选订阅更划算；用量波动大则以一次性购买为主，避免积分闲置到期。 > **⚠ 档位与价格快照说明** > 上表为 **2026-09-04** 的档位快照。档位、价格与积分数量由服务端实时下发，可能随运营调整，实际以[官网首页定价区](https://yeero.ai/zh#pricing)与**账户 → 积分记录 → 购买积分**中的实时展示为准。 ### 免费开始 $0，注册即送 1,000 积分（30 天有效），无需信用卡。 ### 所有档位都包含 - 多模型并行对话 - 知识库与文件管理 - AI 应用快速创作 ## 购买积分 ### 购买入口 - **官网首页定价区** — 未登录时点击任意档位会先引导注册或登录。 - **账户 → 积分记录 → 购买积分** — 已登录时随时可充值；文件页的欠费横幅上也提供"充值积分"入口。 ### 购买步骤 1. 在定价卡片中选择档位，可左右切换查看不同档位，当前聚焦的档位标注"推荐"。 2. 一次性购买可调整**购买数量**，页面实时显示"共 X 积分 · 合计金额"与每积分单价。 3. 选择**支付方式**：Paddle 或微信 / 支付宝，同一档位的微信支付与支付宝可直接切换。 4. 完成支付，积分到账。 ### 信用卡支付（Paddle） 点击订阅或购买后会打开 Paddle 收银台，按提示完成支付即可。支付成功后收银台自动关闭，页面刷新余额与订阅状态。适用于国际信用卡用户，以美元结算。 ### 扫码支付（微信支付 / 支付宝） 选择微信或支付宝后，页面弹出支付二维码，并显示到账积分与应付金额： - 使用对应 App 扫码完成支付。 - **二维码有效期 5 分钟**，页面会自动轮询支付状态，无需手动刷新。 - 支付成功后提示"支付成功，积分已到账"。 - 若提示"支付失败或已超时"或二维码已过期，重新发起一次购买即可。 > **💡 支付后没看到积分？** > 支付回调偶尔有延迟，先刷新页面或重新进入积分记录页。若订单在**账单记录**中仍为"待处理"，说明支付尚未确认，请勿重复付款；持续异常可通过**账户 → 问题反馈**联系我们，并附上订单时间。 ## 订阅管理 订阅以 30 天为一个账单周期，周期开始时发放当月积分。 ### 订阅状态 | 状态 | 含义 | | --- | --- | | **已激活** | 订阅正常运行，按周期自动续费并发放积分。 | | **试用中** | 处于试用期，试用结束前不会扣费。 | | **已暂停** | 订阅暂时停止，恢复后继续按周期计费。 | | **账单逾期** | 续费扣款失败，请及时更新支付方式，否则订阅可能无法继续。 | | **已取消** | 订阅将在当前账单周期结束时停止，在此之前仍可正常使用。 | ### 当前订阅卡片 有订阅时，**账户 → 积分记录**顶部会显示当前订阅卡片，包含套餐名、状态、每月积分、当期账单周期、下次扣费时间与到期时间。 ### 升级档位 1. 在购买积分面板中选择更高档位。 2. 系统会校验当前订阅是否允许该操作：允许则按钮显示"升级"，不允许则直接给出原因提示。 3. 完成支付后提示"升级成功，新订阅周期将从支付时间开始计算"。 ### 取消订阅 1. 前往**账户 → 积分记录**，在当前订阅卡片点击**"取消订阅"**。 2. 确认弹窗会说明三件事：当前账单周期内仍可正常使用、下个周期不会自动续费、以及明确的到期时间。 3. 点击**"确认取消"**；想保留则点击"保留订阅"。 > **💡 取消不等于立刻失效** > 取消后订阅状态变为"已取消"，但你可以继续用到当期周期结束，当月已发放的积分照常使用。周期结束后不再自动续费，当期积分按有效期规则失效。 ## 积分与账单记录 所有积分与账单信息集中在**账户 → 积分记录**（地址为 `/account?tab=credits`）： | 区域 | 内容 | | --- | --- | | 积分余额 | 当前可用积分总数，可手动刷新。 | | 当前订阅 | 有订阅时显示，含状态、下次扣费与到期时间。 | | 购买积分 | 打开定价面板，随时充值或升级。 | | 获取记录 | 每笔积分入账明细，含退款入口。 | | 消耗记录 | 每笔积分扣费明细。 | | 账单记录 | 所有订单与订阅的支付记录。 | ### 账单记录字段 | 字段 | 含义 | | --- | --- | | 类型 | "一次性购买"（单次积分购买）或"订阅"（周期性付款）。 | | 渠道 | Paddle、微信支付或支付宝。 | | 积分 | 该笔记录发放的积分数量。 | | 金额 | 实付金额与币种。 | | 状态 | 订单状态：待处理 / 已完成 / 已退款 / 失败；订阅记录则显示订阅状态。 | | 账单周期 | 订阅记录显示当期起止时间。 | | 时间 | 下单或扣费时间。 | 三个子标签都支持按日期范围筛选（最多回溯一年），列表滚动到底部自动加载更多。 ## 退款 充值购买支持退款，但有明确的时间窗口与计算方式，请先看清楚再提交。 > **⚠ 退款四条硬规则** ### 退款金额怎么算 **退款金额 = 实付金额 × （剩余积分 ÷ 购买总积分）** 举例：花 $10 购买 1,000 积分，已使用 300 积分，剩余 700 积分，则退款 $10 × （700 ÷ 1,000）= **$7.00**，已使用的 300 积分对应的 $3.00 不退。 - 已过期的积分不计入剩余积分。 - 使用优惠券或折扣购买的，按**实付金额**而非原价计算。 - 退款金额按最小货币单位（如美分）四舍五入。 ### 可以申请退款的条件 需同时满足全部条件： 1. 申请在购买后 24 小时内提交。 2. 该笔购买的积分仍有剩余未使用额度。 3. 积分来自直接付费购买，而非促销、赠送或邀请奖励。 4. 账户状态正常，未因违反服务条款被限制或封禁。 5. 该账户此前没有申请过退款。 反之，超过 24 小时、积分已用完、免费获得的积分、账户被封禁、已退过一次款，这五种情况都无法退款。 ### 申请步骤 1. 前往**账户 → 积分记录 → 获取记录**。 2. 找到对应的"充值购买"记录，点击**"申请退款"**。不满足条件时按钮为禁用状态。 3. 核对退款详情：订单信息、购买积分、已使用、剩余积分、预计退款与"原路退回"说明。 4. 点击**"确认退款"**提交，提交成功后显示"退款申请已提交，正在处理中"。 提交后该记录会显示**"退款处理中"**，完成后显示**"已退款"**；被收回的剩余积分在消耗记录中以"退还积分"体现。 ### 到账时间与状态 | 阶段 | 时长 | | --- | --- | | 平台内部审核 | 1-3 个工作日 | | 支付渠道结算到账 | 5-15 个工作日，因支付渠道与地区而异 | 退款状态分为**处理中**（审核中）、**已完成**（已通过并提交支付渠道）与**失败**三种。退款一律**原路退回**至原支付方式，不支持更换渠道或转为账户余额。 完整条款以[退款政策](https://yeero.ai/policy/refund/)为准。 ## 邀请奖励 入口在**账户 → 邀请有礼**： - **你的邀请码** — 点击"复制邀请码"分享给好友。 - **邀请链接** — 分享专属链接，好友通过链接注册时邀请码自动生效。 - **10% 奖励** — 好友注册时你获得其注册积分的 10%，好友每次购买积分时你再获得其购买积分的 10%。 - **无上限** — 邀请越多奖励越多，且邀请获得的积分永不过期。 - **邀请数据** — 页面显示邀请人数、累计获得奖励，以及已邀请用户列表（滚动加载更多）。 ### 邀请规则 1. 每位用户拥有专属邀请码，分享给好友即可邀请。 2. 好友通过你的邀请码注册后，你获得其注册积分的 10% 奖励。 3. 好友每次购买积分时，你获得其购买积分的 10% 奖励。 4. 邀请奖励无上限，邀请越多奖励越多。 5. 通过邀请获取的积分永不过期。 6. 严禁刷号行为，违者将直接封号。 邀请获得的积分在获取记录中标记为"邀请奖励"，但不属于可退款范围。 ## 还有疑问 - [常见问题](https://yeero.ai/docs/zh/faq.html) — 积分、模型、隐私与故障排查的高频问答。 - **账户 → 问题反馈** — 提交故障报告或功能建议，可附截图，并能在同一页跟踪处理状态。 - [退款政策](https://yeero.ai/policy/refund/)与[服务条款](https://yeero.ai/policy/terms/) — 完整法务条款。 --- # 常见问题 *关于使用 YeeroAI 的常见问题解答。* > **⚡ 一句话摘要** > 关于 YeeroAI 最常见问题的直接答案——账号、积分、模型、核心功能、桌面端与 CLI、隐私、存储与故障排查，一页看全。 ## 账户与登录 ### 如何注册？ 访问 yeero.ai，使用邮箱（验证码登录）、Google 或 GitHub 注册。你将立即获得 **1,000** 免费积分（30 天有效期）。 ### 可以使用测试账户吗？ 可以，在登录页面点击"试用测试账户"。测试账户功能有限——部分操作（如支付）受限，适合先体验再决定是否注册。 ### 如何更改邮箱？ 前往 账户 → 账号安全，点击邮箱旁边的"绑定"或"切换"。你需要先通过当前登录方式完成身份验证。 ### 可以绑定多种登录方式吗？ 可以，你可以将邮箱、Google 和 GitHub 绑定到同一个账户。我们建议绑定多种方式以提高安全性，但必须至少保留一种登录方式。 ### 如何退出登录？ 前往 账户 → 登出账号，点击"登出"退出当前设备，或点击"登出全部设备"强制下线所有已登录设备。 ### 邮箱登录收不到验证码怎么办？ 验证码邮件可能在垃圾邮件箱中，请先检查。若仍未收到，等待倒计时结束后点击"重新发送"。如持续无法收到，请通过 账户 → 问题反馈 联系我们。 ### 如何注销账户？ 出于数据安全考虑，账户注销需人工处理。请前往 账户 → 问题反馈 提交注销申请（标题注明"账户注销"），我们将在 30 天内删除或匿名化你的全部数据。详见 [隐私政策](https://yeero.ai/policy/privacy/)。 ## 积分与计费 ### 我能获得多少免费积分？ 注册即送 **1,000** 积分（30 天有效期）。邀请好友还可获得额外积分——邀请积分永不过期。 ### 积分如何消耗？ 积分用于 AI 对话、记忆生成与匹配、AI 应用生成与讨论、Agent 与云端任务，以及超额存储。消耗金额取决于模型和输出长度，每条 AI 响应上都会实时显示确切消耗。完整清单见 [消耗积分](https://yeero.ai/docs/zh/billing.html#spending)。 ### 积分会过期吗？ 一次性购买的积分永不过期。订阅积分在订阅结束后可能过期。注册赠送的 1,000 积分有效期为 30 天，其他赠送积分的有效期可在 账户 → 积分记录 → 获取记录 的"过期时间"列查看。详见 [积分有效期](https://yeero.ai/docs/zh/billing.html#expiry)。 ### 可以退款吗？ 可以。前往 账户 → 积分记录 → 获取记录，在充值购买记录中点击"申请退款"。退款需在购买后 **24 小时**内发起，按剩余积分比例计算，每个账户仅限一次，提交后不可撤销。内部审核 1-3 个工作日，原路退回到账另需 5-15 个工作日。详见 [退款政策](https://yeero.ai/policy/refund/)与 [退款](https://yeero.ai/docs/zh/billing.html#refunds)。 ### 支持哪些支付方式？ Paddle（国际信用卡）和微信支付 / 支付宝（中国用户）。所有支付均通过持牌第三方处理，YeeroAI 不存储完整卡号。 ### 订阅和一次性购买有什么区别？ 订阅按月提供固定积分，适合持续使用；一次性购买按需付费，积分永不过期。两者可叠加使用，订阅可随时取消，取消后当前计费周期内仍可继续使用。 ### 在哪里查看积分消耗明细？ 前往 账户 → 积分记录 → 消耗记录，查看每笔扣费（对话、记忆、应用、存储）。同一页面的 账单记录 子标签中可查看所有订单和订阅历史。 ### 邀请奖励如何计算？ 你赚取好友注册积分的 **10%**，以及他们每次购买积分的 **10%**。邀请奖励无上限，邀请积分永不过期。在 账户 → 邀请有礼 中获取你的专属邀请码和链接。详见 [邀请奖励](https://yeero.ai/docs/zh/billing.html#invitations)。 ## AI 模型 ### 有多少模型可用？ **300+** 个模型，来自 10+ 个提供商，包括 OpenAI、Anthropic、Google、Meta、Mistral 等，全部通过 OpenRouter 统一接入。 ### 有免费模型吗？ 有，在模型选择器中切换"仅免费"即可查看免费模型。免费模型每次调用消耗 **1** 积分。 ### 可以同时使用多个模型吗？ 可以，在同一对话中最多选择 **10** 个模型进行并行对比，回答并排显示，方便横向评估。 ### 支持哪些模态？ 文本、图片、音频、视频和文件输入 / 输出，具体取决于所选模型的能力。 ### 如何选择合适的模型？ 日常对话可选通用模型（如 GPT、Claude）；代码生成可选擅长编程的模型；需要长上下文时关注模型的上下文长度。模型详情页列出了能力标签、上下文长度和积分单价，可据此比较。 ### 某个模型不可用怎么办？ 模型偶尔会因上游供应商维护而暂时不可用。你可以切换到其他模型继续对话；如果问题持续，请通过 账户 → 问题反馈 告知我们。 ## 核心功能 ### 什么是分支思维？ 分支思维为每个对话提供 Git 式分支：从任意消息派生新路径探索不同方向，合并保留最佳思路，随时回溯到任意节点——主线对话永远不受影响。详见 [分支思维](https://yeero.ai/docs/zh/branching.html)。 ### 什么是追问？ 选中 AI 回复中的某段文字，一键发起追问。追问在独立线程中展开，支持层层深入，主对话的上下文保持完整不受影响。详见 [追问](https://yeero.ai/docs/zh/followup.html)。 ### AI 应用是什么？ 用自然语言描述想法，AI 即可生成可运行的 HTML 或 Python 应用。支持补丁式增量迭代、完整版本历史、实时预览和内置讨论协作。详见 [AI 应用](https://yeero.ai/docs/zh/ai-apps.html)。 ### 知识库如何工作？ AI 自动从你的对话中提取关键知识，构建向量索引，并在未来对话中注入相关上下文。你可以按标签浏览、搜索和管理知识条目。详见 [知识库](https://yeero.ai/docs/zh/knowledge-base.html)。 ### AI 记忆是什么？ 记忆是 AI 从对话中提炼的持久化知识条目，会在后续对话中自动匹配并注入相关上下文，让 AI 越用越懂你。你可以在 设置 → 记忆 中查看、编辑和删除记忆。 ## 桌面端与 CLI ### 有桌面客户端吗？ 有。Yeero Desktop 基于 Tauri 构建，支持 macOS（Apple Silicon / Intel）、Windows（x64 / ARM64）和 Linux（AppImage / deb），提供原生窗口、系统通知和应用内自动更新。详见 [桌面端](https://yeero.ai/docs/zh/desktop.html)。 ### Launcher 是什么？ Launcher 是桌面端的全局唤起器，按 `Alt`+`Space` 即可呼出。内置 16 个处理器，覆盖计算、启动、搜索与 AI 对话。详见 [Launcher](https://yeero.ai/docs/zh/launcher.html)。 ### 桌面端支持截图录屏吗？ 支持。桌面端提供窗口智能识别、长截图拼接、麦克风与摄像头画中画录制，截图后可直接进入图片编辑器标注。详见 [截图与录屏](https://yeero.ai/docs/zh/capture.html)。 ### 有命令行工具吗？ 有。yeero CLI 一条命令安装：macOS / Linux 运行 `curl -fsSL https://yeero.ai/cli/install.sh | sh`，Windows 运行 `irm https://yeero.ai/cli/install.ps1 | iex`。详见 [CLI 快速入门](https://yeero.ai/docs/zh/cli-getting-started.html)。 ### CLI 和网页端数据互通吗？ 互通。CLI 登录后与网页端、桌面端共用同一账号，对话历史实时同步。 ## 隐私与安全 ### 我的数据是加密的吗？ 是的。所有数据在传输过程中使用 **TLS 1.2+** 加密，存储时使用 **AES-256** 加密。你的私有数据绝不会被出售或与任何第三方共享。 ### 可以删除我的数据吗？ 可以，你可以随时删除单个对话、消息、文件和知识条目。注销账户后，所有数据将在 **30 天**内被删除或匿名化。删除是永久性的，无法撤销。 ### 关于 AI 训练呢？ YeeroAI 不会将你的数据用于 AI 模型训练。但你的查询会发送至底层模型提供商（如 OpenAI、Anthropic、OpenRouter）以生成回答，这些数据受相应提供商隐私政策约束。详见 [隐私政策](https://yeero.ai/policy/privacy/)。 ### 端到端加密密钥是什么？ 在 设置 → 端到端加密密钥 中可配置通用 E2E 密钥，用于加密同步场景（如剪贴板跨设备同步）。密钥仅保存在你本机的系统密钥链中，留空表示不开启。要在多台设备间解密已同步内容，需在每台设备配置相同密钥。 ### 数据存储在哪里？ YeeroAI 将你的数据存储在通过 SOC 2 Type II 和 ISO 27001 认证的云基础设施上，传输与存储全程加密，并受严格的访问控制保护。详见 [隐私政策](https://yeero.ai/policy/privacy/)。 ## 存储与文件 ### 我有多少存储空间？ **1 GB** 免费存储。超出免费额度后，超额存储会产生每日积分费用。 ### 超出存储会怎样？ 你会看到警告横幅。如果存储逾期超过 **7 天**且积分余额不足，系统会自动删除最旧的文件，直到你回到 1 GB 免费额度内。 ### 可以上传什么文件类型？ 支持任何格式，单个文件最大 **512 MB**。图片、视频、音频和文档均可在线预览。 ### 我上传的文件会被 AI 读取吗？ 仅当你主动将文件附加到对话或应用中时，AI 才会读取该文件内容。未附加的文件不会被自动分析。文件会建立语义索引以支持搜索，但这与对话上下文注入是分开的。 ### 如何批量清理文件？ 文件页支持鼠标拖拽矩形框选，圈选后可批量下载、删除或移动。你也可以在 清理规则 中设置自动删除条件（按类型、来源、大小、路径、时间），规则每天凌晨 2 点自动运行。 ## 故障排查与支持 ### 支持哪些浏览器？ YeeroAI 在现代浏览器上运行最佳：Chrome、Firefox、Safari 和 Edge。建议使用最新版本以获得完整功能（如 Service Worker、SSE 流式传输）。 ### YeeroAI 在手机上能用吗？ 可以，Web 界面完全响应式，可在移动浏览器上使用。原生移动端应用正在开发中，桌面端已提供原生体验。 ### AI 生成失败怎么办？ 先尝试刷新页面。如果问题持续，检查积分余额是否充足，或切换到其他模型。后台流式传输会在网络中断后自动恢复，无需手动重发。 ### 流式响应中断了怎么办？ YeeroAI 使用 SSE 流式传输并支持断线重连。网络恢复后会自动续传，已生成的内容不会丢失。如果长时间无响应，可点击"停止"后重新发送。 ### 云同步不生效怎么办？ 确认 设置 → 云同步 中对应开关已开启。同步采用防抖推送（约 1 秒），切换设备后稍等片刻或刷新页面即可拉取最新数据。部分临时 UI 状态不参与同步。 ### 如何反馈问题或建议？ 前往 账户 → 问题反馈 提交问题、建议或投诉，我们会跟进处理。涉及退款申诉、隐私或法律事务，也请通过同一入口提交并注明类型。 > **💬 没找到答案？** > 前往 账户 → 问题反馈 直接告诉我们，或浏览完整文档。 --- Copyright (c) 2026 YeeroAI. All rights reserved. This content may be cited with attribution and a link to https://yeero.ai, but may not be republished in full without permission.