diff --git a/src/os-skills/README.md b/src/os-skills/README.md
index ee6fa8d..35f6ee5 100644
--- a/src/os-skills/README.md
+++ b/src/os-skills/README.md
@@ -45,6 +45,9 @@ A curated collection of operational skills for AI Agents, covering system admini
 ### Security
 - **alinux-cve-query** — Query Alibaba Cloud Linux CVE vulnerability info
 
+### Other
+- **cosh-guide** - Copilot Shell user guide
+
 ## Skill Format
 
 Each skill lives in its own directory with at least a `SKILL.md` file:
diff --git a/src/os-skills/others/cosh-guide/SKILL.md b/src/os-skills/others/cosh-guide/SKILL.md
new file mode 100644
index 0000000..82ebbe8
--- /dev/null
+++ b/src/os-skills/others/cosh-guide/SKILL.md
@@ -0,0 +1,45 @@
+---
+name: cosh-guide
+description: Copilot Shell (cosh) CLI 用户指南和帮助文档。用于回答关于 Copilot Shell 的使用、命令、配置、功能特性等问题。
+  当用户询问 Copilot Shell 的使用方法、命令参考、配置方式、功能说明时使用此 skill。
+---
+
+# Copilot Shell (COSH) 用户指南
+
+你是 Copilot Shell (COSH) CLI 工具的用户帮助助手。当用户询问 Copilot Shell 的使用问题时，根据问题类型参考对应的文档来回答。
+
+## 文档索引
+
+根据用户问题的类别，参考以下对应的参考文档：
+
+### 命令参考
+- 斜杠命令 (`/`)、@文件引用、`!`shell 命令、自定义命令、快捷键
+- 参考: [commands.md](reference/commands.md)
+
+### 功能特性
+- 审批模式 (Plan/Default/Auto-Edit/YOLO)
+- Headless 模式 (非交互式/脚本化)
+- MCP (Model Context Protocol) 外部工具集成
+- Agent Skills (实验性扩展能力)
+- Subagents (子代理/专业化AI助手)
+- Checkpointing (检查点/状态恢复)
+- LSP (语言服务器协议，代码智能)
+- Token 缓存与成本优化
+- 国际化与语言设置
+- 参考: [features.md](reference/features.md)
+
+### 配置
+- settings.json 配置层级与选项
+- 认证 (OAuth / API Key)
+- 主题设置与自定义
+- 可信文件夹
+- 命令行参数
+- 环境变量
+- 参考: [configuration.md](reference/configuration.md)
+
+## 回答规范
+
+1. 根据用户问题判断属于哪个类别，读取对应的参考文档
+2. 用简洁清晰的语言回答，提供具体的命令或配置示例
+3. 如果问题涉及多个类别，综合多个参考文档的信息
+4. 对于不确定的信息，建议用户使用 `/help` 命令或查看官方文档
diff --git a/src/os-skills/others/cosh-guide/reference/commands.md b/src/os-skills/others/cosh-guide/reference/commands.md
new file mode 100644
index 0000000..80deb70
--- /dev/null
+++ b/src/os-skills/others/cosh-guide/reference/commands.md
@@ -0,0 +1,191 @@
+# Commands Reference
+
+Qwen Code commands use three prefixes:
+
+| Prefix | Function | Example |
+|---|---|---|
+| `/` (Slash) | Meta-level control of Qwen Code | `/help`, `/settings` |
+| `@` (At) | Inject local file content into conversation | `@src/main.py explain this` |
+| `!` (Exclamation) | Direct system shell interaction | `!git status`, `!ls` |
+
+---
+
+## 1. Slash Commands (`/`)
+
+### 1.1 Session and Project Management
+
+| Command | Description | Usage |
+|---|---|---|
+| `/init` | Analyze directory and create initial context file | `/init` |
+| `/summary` | Generate project summary from conversation | `/summary` |
+| `/compress` | Replace chat history with summary to save tokens | `/compress` |
+| `/resume` | Resume a previous conversation session | `/resume` |
+| `/restore` | Restore files to state before tool execution | `/restore` or `/restore <ID>` |
+
+### 1.2 Interface and Workspace Control
+
+| Command | Description | Usage |
+|---|---|---|
+| `/clear` | Clear terminal screen | `/clear` (Ctrl+L) |
+| `/theme` | Change visual theme | `/theme` |
+| `/vim` | Toggle Vim editing mode | `/vim` |
+| `/directory` | Manage multi-directory workspace | `/dir add ./src,./tests` |
+| `/editor` | Select supported editor | `/editor` |
+
+### 1.3 Language Settings
+
+| Command | Description | Usage |
+|---|---|---|
+| `/language` | View/change language settings | `/language` |
+| `/language ui [lang]` | Set UI language | `/language ui zh-CN` |
+| `/language output [lang]` | Set LLM output language | `/language output Chinese` |
+
+Built-in UI languages: `zh-CN`, `en-US`, `ru-RU`, `de-DE`, `ja-JP`
+Aliases: `zh`, `en`, `ru`, `de`, `ja`
+
+### 1.4 Tool and Model Management
+
+| Command | Description | Usage |
+|---|---|---|
+| `/mcp` | List configured MCP servers/tools | `/mcp`, `/mcp desc` |
+| `/tools` | Display available tool list | `/tools`, `/tools desc` |
+| `/skills` | List/run available skills (experimental) | `/skills`, `/skills <name>` |
+| `/approval-mode` | Change approval mode | `/approval-mode <mode> --project` |
+| `/model` | Switch model in current session | `/model` |
+| `/extensions` | List active extensions | `/extensions` |
+| `/memory` | Manage AI instruction context | `/memory add Important Info` |
+
+Approval modes: `plan` (read-only), `default` (manual approval), `auto-edit` (auto-approve edits), `yolo` (auto-approve all)
+
+### 1.5 Information, Settings, and Help
+
+| Command | Description | Usage |
+|---|---|---|
+| `/help` | Display help | `/help` or `/?` |
+| `/about` | Display version info | `/about` |
+| `/stats` | Session statistics | `/stats` |
+| `/settings` | Open settings editor | `/settings` |
+| `/auth` | Change authentication | `/auth` |
+| `/bug` | Submit issue | `/bug Button unresponsive` |
+| `/copy` | Copy last output to clipboard | `/copy` |
+| `/quit` | Exit Qwen Code | `/quit` or `/exit` |
+
+---
+
+## 2. @ Commands (File References)
+
+| Format | Description | Example |
+|---|---|---|
+| `@<file path>` | Inject file content | `@src/main.py Please explain this` |
+| `@<directory>` | Read all text files recursively | `@docs/ Summarize this` |
+| `@server: resource` | Reference MCP resources | `@github: repos/owner/repo/issues` |
+
+Notes:
+- Spaces in paths need backslash escaping: `@My\ Documents/file.txt`
+- File paths can be relative or absolute
+- `@` file references auto-load `QWEN.md` from the file's directory and parents
+- Multiple files: `@file1.js and @file2.js`
+
+---
+
+## 3. Exclamation Commands (`!`) - Shell
+
+| Format | Description | Example |
+|---|---|---|
+| `!<command>` | Execute in sub-shell | `!ls -la`, `!git status` |
+| Standalone `!` | Toggle shell mode (all input as commands) | `!` (enter) -> commands -> `!` (exit) |
+
+Environment: Commands via `!` set `QWEN_CODE=1`.
+
+---
+
+## 4. Custom Commands
+
+Save frequently used prompts as shortcut commands.
+
+### Storage Locations
+
+| Type | Location | Priority |
+|---|---|---|
+| Global | `~/.qwen/commands/` | Low |
+| Project | `<project>/.qwen/commands/` | High (overrides global) |
+
+### Naming
+
+File path maps to command name:
+- `~/.qwen/commands/test.md` -> `/test`
+- `<project>/.qwen/commands/git/commit.md` -> `/git:commit`
+
+### Markdown Format (Recommended)
+
+```markdown
+---
+description: Optional description (shown in /help)
+---
+
+Your prompt content here.
+Use {{args}} for parameter injection.
+```
+
+### Dynamic Content Injection
+
+| Type | Syntax | Purpose |
+|---|---|---|
+| File Content | `@{file path}` | Inject static reference files |
+| Shell Commands | `!{command}` | Inject execution results (requires confirmation) |
+| Parameters | `{{args}}` | Inject user parameters |
+
+### Example: Git Commit Command
+
+````markdown
+---
+description: Generate Commit message based on staged changes
+---
+
+Please generate a Commit message based on the following diff:
+
+```diff
+!{git diff --staged}
+```
+````
+
+---
+
+## 5. Keyboard Shortcuts
+
+### General
+
+| Shortcut | Description |
+|---|---|
+| `Esc` | Close dialogs/suggestions |
+| `Ctrl+C` | Cancel request / clear input (press twice to exit) |
+| `Ctrl+D` | Exit if input empty (press twice to confirm) |
+| `Ctrl+L` | Clear screen |
+| `Ctrl+O` | Toggle debug console |
+| `Ctrl+S` | Allow long responses to print fully |
+| `Ctrl+T` | Toggle tool descriptions |
+| `Shift+Tab` | Cycle approval modes |
+
+### Input Prompt
+
+| Shortcut | Description |
+|---|---|
+| `!` | Toggle shell mode (empty input) |
+| `?` | Toggle shortcuts display (empty input) |
+| `\` + Enter | Insert newline |
+| `Tab` | Autocomplete suggestion |
+| Up/Down Arrow | Navigate input history |
+| `Ctrl+R` | Reverse search history |
+| `Ctrl+V` | Paste clipboard (supports images) |
+| `Ctrl+X` / `Meta+Enter` | Open in external editor |
+| `Ctrl+A` / `Home` | Beginning of line |
+| `Ctrl+E` / `End` | End of line |
+| `Ctrl+K` | Delete to end of line |
+| `Ctrl+U` | Delete to beginning of line |
+| `Ctrl+W` | Delete word left |
+
+### IDE Integration
+
+| Shortcut | Description |
+|---|---|
+| `Ctrl+G` | See context from IDE |
diff --git a/src/os-skills/others/cosh-guide/reference/configuration.md b/src/os-skills/others/cosh-guide/reference/configuration.md
new file mode 100644
index 0000000..e72e06b
--- /dev/null
+++ b/src/os-skills/others/cosh-guide/reference/configuration.md
@@ -0,0 +1,249 @@
+# Configuration Reference
+
+## Configuration Layers (Precedence)
+
+| Level | Source | Description |
+|---|---|---|
+| 1 | Default values | Hardcoded defaults |
+| 2 | User settings | `~/.copilot/settings.json` |
+| 3 | Project settings | `.copilot/settings.json` |
+
+---
+
+## Authentication
+### Switch Method
+
+```
+/auth
+```
+
+---
+
+## Settings (settings.json)
+
+### Key Settings
+
+#### general
+
+| Setting | Type | Default | Description |
+|---|---|---|---|
+| `general.preferredEditor` | string | undefined | Preferred editor |
+| `general.vimMode` | boolean | false | Enable Vim keybindings |
+| `general.disableAutoUpdate` | boolean | false | Disable auto-updates |
+| `general.gitCoAuthor` | boolean | true | Add Co-authored-by to commits |
+| `general.checkpointing.enabled` | boolean | false | Enable checkpointing |
+
+#### model
+
+| Setting | Type | Default | Description |
+|---|---|---|---|
+| `model.name` | string | undefined | Model to use |
+| `model.maxSessionTurns` | number | -1 | Max turns (-1 = unlimited) |
+| `model.chatCompression.contextPercentageThreshold` | number | 0.7 | Compression threshold (0-1) |
+| `model.generationConfig` | object | undefined | Advanced overrides (timeout, maxRetries, samplingParams, etc.) |
+
+#### ui
+
+| Setting | Type | Default | Description |
+|---|---|---|---|
+| `ui.theme` | string | undefined | Color theme |
+| `ui.hideTips` | boolean | false | Hide tips |
+| `ui.showLineNumbers` | boolean | true | Line numbers in code blocks |
+
+#### tools
+
+| Setting | Type | Default | Description |
+|---|---|---|---|
+| `tools.approvalMode` | string | "default" | Default approval mode |
+| `tools.sandbox` | bool/string | undefined | Sandbox execution |
+| `tools.core` | array | undefined | Allowlist of built-in tools |
+| `tools.exclude` | array | undefined | Tools to exclude |
+| `tools.allowed` | array | undefined | Tools bypassing confirmation |
+
+#### context
+
+| Setting | Type | Default | Description |
+|---|---|---|---|
+| `context.fileName` | string/array | undefined | Context file name(s) |
+| `context.includeDirectories` | array | [] | Extra directories |
+| `context.fileFiltering.respectGitIgnore` | boolean | true | Respect .gitignore |
+| `context.fileFiltering.respectQwenIgnore` | boolean | true | Respect .qwenignore |
+
+#### privacy
+
+| Setting | Type | Default | Description |
+|---|---|---|---|
+| `privacy.usageStatisticsEnabled` | boolean | true | Usage statistics |
+
+### Example settings.json
+
+```json
+{
+  "general": { "vimMode": true, "preferredEditor": "code" },
+  "ui": { "theme": "GitHub" },
+  "tools": { "approvalMode": "auto-edit" },
+  "model": { "name": "qwen3-coder-plus" },
+  "context": {
+    "fileName": ["QWEN.md"],
+    "includeDirectories": ["path/to/dir1"]
+  }
+}
+```
+
+---
+
+## Themes
+
+### Built-in Themes
+
+- **Dark**: ANSI, Atom One, Ayu, Default, Dracula, GitHub
+- **Light**: ANSI Light, Ayu Light, Default Light, GitHub Light, Google Code, Xcode
+
+### Change Theme
+
+```
+/theme
+```
+
+### Custom Themes
+
+In `settings.json`:
+
+```json
+{
+  "ui": {
+    "customThemes": {
+      "MyTheme": {
+        "name": "MyTheme",
+        "type": "custom",
+        "Background": "#181818",
+        "Foreground": "#F8F8F2",
+        "LightBlue": "#82AAFF",
+        "AccentBlue": "#61AFEF",
+        "AccentPurple": "#BD93F9",
+        "AccentCyan": "#8BE9FD",
+        "AccentGreen": "#50FA7B",
+        "AccentYellow": "#F1FA8C",
+        "AccentRed": "#FF5555",
+        "Comment": "#6272A4",
+        "Gray": "#ABB2BF"
+      }
+    }
+  }
+}
+```
+
+Load from file: `"ui": { "theme": "/path/to/theme.json" }`
+
+---
+
+## .qwenignore
+
+Exclude files from Qwen Code tools (similar to `.gitignore`).
+
+Create `.qwenignore` in project root:
+
+```
+# Exclude directories
+/packages/
+
+# Exclude specific files
+apikeys.txt
+
+# Wildcards
+*.md
+
+# Negate (include back)
+!README.md
+```
+
+Restart session after changes.
+
+---
+
+## Trusted Folders
+
+Security feature controlling which projects can use full Qwen Code capabilities.
+
+### Enable
+
+```json
+{ "security": { "folderTrust": { "enabled": true } } }
+```
+
+### Trust Dialog Options
+
+- **Trust folder**: Full trust for current folder
+- **Trust parent**: Trust parent directory and all subdirectories
+- **Don't trust**: Restricted "safe mode"
+
+### Untrusted Restrictions
+
+1. Project `.qwen/settings.json` ignored
+2. Project `.env` files ignored
+3. Extension management restricted
+4. Tool auto-acceptance disabled
+5. Automatic memory loading disabled
+
+### Manage
+
+- `/permissions` command to change current folder trust
+- View all rules: `~/.qwen/trustedFolders.json`
+
+---
+
+## Context Files (QWEN.md)
+
+Hierarchical instructional context for the AI.
+
+### Loading Order
+
+1. `~/.qwen/QWEN.md` (global)
+2. Project root and parent directories (project-specific)
+
+### Commands
+
+- `/memory refresh` - Reload all context files
+- `/memory show` - Display current loaded context
+- `/memory add <text>` - Add to memory
+
+### Import Syntax
+
+Use `@path/to/file.md` in context files to import other Markdown files.
+
+---
+
+## Command-Line Arguments
+
+| Argument | Alias | Description |
+|---|---|---|
+| `--model` | `-m` | Specify model |
+| `--prompt` | `-p` | Headless mode |
+| `--output-format` | `-o` | Output format |
+| `--sandbox` | `-s` | Enable sandbox |
+| `--debug` | `-d` | Debug mode |
+| `--all-files` | `-a` | Include all files |
+| `--yolo` | | Auto-approve all |
+| `--approval-mode` | | Set approval mode |
+| `--continue` | | Resume recent session |
+| `--resume` | | Resume specific session |
+| `--include-directories` | | Add extra directories |
+| `--experimental-skills` | | Enable skills |
+| `--experimental-lsp` | | Enable LSP |
+| `--extensions` | `-e` | Specify extensions |
+| `--proxy` | | Set proxy |
+| `--checkpointing` | | Enable checkpointing |
+
+---
+
+## Environment Variables
+
+| Variable | Description |
+|---|---|
+| `OPENAI_API_KEY` | API key for OpenAI-compatible provider |
+| `OPENAI_BASE_URL` | Custom API base URL |
+| `OPENAI_MODEL` | Model to use |
+| `QWEN_CODE_LANG` | Override UI language detection |
+| `TAVILY_API_KEY` | Tavily web search API key |
+| `NO_COLOR` | Disable color output |
+| `DEBUG` / `DEBUG_MODE` | Enable debug logging |
diff --git a/src/os-skills/others/cosh-guide/reference/features.md b/src/os-skills/others/cosh-guide/reference/features.md
new file mode 100644
index 0000000..bee9dda
--- /dev/null
+++ b/src/os-skills/others/cosh-guide/reference/features.md
@@ -0,0 +1,345 @@
+# Features Reference
+
+## Approval Mode
+
+Four permission modes controlling how AI interacts with your code:
+
+| Mode | File Editing | Shell Commands | Best For | Risk |
+|---|---|---|---|---|
+| **Plan** | Read-only | Not executed | Code exploration, planning | Lowest |
+| **Default** | Manual approval | Manual approval | New codebases, critical systems | Low |
+| **Auto-Edit** | Auto-approved | Manual approval | Daily development, refactoring | Medium |
+| **YOLO** | Auto-approved | Auto-approved | Trusted projects, CI/CD | Highest |
+
+### Switching Modes
+
+- **During session**: `Shift+Tab` to cycle
+- **Command**: `/approval-mode <plan|default|auto-edit|yolo>`
+- **With scope**: `/approval-mode yolo --project` or `--user`
+- **Config**: `"permissions": { "defaultMode": "auto-edit" }` in `settings.json`
+
+### Plan Mode for Complex Refactors
+
+```bash
+/approval-mode plan
+```
+```
+I need to refactor our auth system to use OAuth2. Create a detailed migration plan.
+```
+
+---
+
+## Headless Mode
+
+Run Qwen Code programmatically without interactive UI.
+
+### Basic Usage
+
+```bash
+# Direct prompt
+qwen --prompt "What is machine learning?"
+qwen -p "query"
+
+# Stdin input
+echo "Explain this code" | qwen
+
+# File input
+cat README.md | qwen -p "Summarize this documentation"
+```
+
+### Output Formats
+
+| Format | Flag | Description |
+|---|---|---|
+| Text (default) | `--output-format text` | Human-readable |
+| JSON | `--output-format json` | Structured, buffered |
+| Stream-JSON | `--output-format stream-json` | Real-time streaming |
+
+### Resume Sessions
+
+```bash
+qwen --continue -p "Run the tests again"
+qwen --resume <session-id> -p "Apply the refactor"
+```
+
+### Key Options
+
+| Option | Description |
+|---|---|
+| `--prompt`, `-p` | Run headless |
+| `--output-format`, `-o` | Output format (text/json/stream-json) |
+| `--yolo`, `-y` | Auto-approve all |
+| `--approval-mode` | Set approval mode |
+| `--all-files`, `-a` | Include all files in context |
+| `--include-directories` | Include extra directories |
+| `--continue` | Resume most recent session |
+| `--resume [id]` | Resume specific session |
+
+---
+
+## MCP (Model Context Protocol)
+
+Connect Qwen Code to external tools and data sources.
+
+### Quick Start
+
+```bash
+# Add HTTP server
+qwen mcp add --transport http my-server http://localhost:3000/mcp
+
+# Add stdio server
+qwen mcp add pythonTools python -m my_mcp_server
+
+# List servers
+qwen mcp list
+
+# Remove server
+qwen mcp remove my-server
+```
+
+### Transport Types
+
+| Transport | When to Use | Config Field |
+|---|---|---|
+| `http` | Remote services (recommended) | `httpUrl` |
+| `sse` | Legacy SSE servers | `url` |
+| `stdio` | Local processes | `command`, `args` |
+
+### Configuration via settings.json
+
+```json
+{
+  "mcpServers": {
+    "myServer": {
+      "command": "python",
+      "args": ["-m", "my_mcp_server"],
+      "env": { "API_KEY": "$MY_API_KEY" },
+      "timeout": 15000
+    }
+  }
+}
+```
+
+### Scopes
+
+- **Project** (default): `.copilot/settings.json`
+- **User**: `~/.copilot/settings.json` (use `--scope user`)
+
+### Safety
+
+- `trust: true` bypasses confirmations (use sparingly)
+- `includeTools` / `excludeTools` to filter tools per server
+- Global allow/deny: `mcp.allowed` / `mcp.excluded` in settings
+
+---
+
+## Agent Skills
+Modular capabilities extending Cosh.
+
+### Create a Skill
+
+Skills are directories containing `SKILL.md`:
+
+- **Personal**: `~/.copilot/skills/<skill-name>/SKILL.md`
+- **Project**: `.copilot/skills/<skill-name>/SKILL.md`
+
+```yaml
+---
+name: your-skill-name
+description: Brief description of what this Skill does and when to use it
+---
+
+# Your Skill Name
+
+## Instructions
+Step-by-step guidance.
+
+## Examples
+Concrete examples.
+```
+
+### Invoke
+
+Skills are auto-invoked by the model based on your request. Explicit invocation:
+
+```bash
+/skills <skill-name>
+```
+
+### Supporting Files
+
+```
+my-skill/
+  SKILL.md (required)
+  reference.md (optional)
+  scripts/helper.py (optional)
+  templates/template.txt (optional)
+```
+
+---
+
+## Subagents
+
+Specialized AI assistants for specific task types.
+
+### Management
+
+```
+/agents create    # Create new subagent
+/agents manage    # View/manage existing
+```
+
+### Storage
+
+- **Project**: `.copilot/agents/` (highest precedence)
+- **User**: `~/.copilot/agents/`
+- **Extension**: Provided by installed extensions
+
+### Configuration Format
+
+```markdown
+---
+name: agent-name
+description: Brief description
+tools:
+  - read_file
+  - write_file
+  - run_shell_command
+---
+
+System prompt content goes here.
+Use ${variable} templating for dynamic content.
+```
+
+### Invocation
+
+- **Automatic**: AI delegates based on task matching
+- **Explicit**: "Let the testing-expert subagent create unit tests for the payment module"
+
+### Best Practices
+
+- Single responsibility per agent
+- Specific expertise areas
+- Clear, actionable descriptions
+- Limit tool access to what's needed
+
+---
+
+## Checkpointing
+
+Auto-saves project state before AI file modifications.
+
+### Enable
+
+```bash
+qwen --checkpointing
+```
+
+Or in `settings.json`:
+```json
+{ "general": { "checkpointing": { "enabled": true } } }
+```
+
+### How It Works
+
+Before file modifications, creates:
+1. Git snapshot in shadow repository (`~/.qwen/history/<project_hash>`)
+2. Conversation history backup
+3. Tool call record
+
+### Restore
+
+```
+/restore           # List available checkpoints
+/restore <file>    # Restore specific checkpoint
+```
+
+---
+
+## LSP (Language Server Protocol) - Experimental
+
+Code intelligence features via language servers.
+
+### Enable
+
+```bash
+qwen --experimental-lsp
+```
+
+### Supported Languages
+
+| Language | Server | Install |
+|---|---|---|
+| TypeScript/JS | typescript-language-server | `npm install -g typescript-language-server typescript` |
+| Python | pylsp | `pip install python-lsp-server` |
+| Go | gopls | `go install golang.org/x/tools/gopls@latest` |
+| Rust | rust-analyzer | See official docs |
+
+### Configuration (.lsp.json)
+
+```json
+{
+  "typescript": {
+    "command": "typescript-language-server",
+    "args": ["--stdio"],
+    "extensionToLanguage": {
+      ".ts": "typescript", ".tsx": "typescriptreact",
+      ".js": "javascript", ".jsx": "javascriptreact"
+    }
+  }
+}
+```
+
+### Available Operations
+
+- `goToDefinition`, `findReferences`, `goToImplementation`
+- `hover`, `documentSymbol`, `workspaceSymbol`
+- `prepareCallHierarchy`, `incomingCalls`, `outgoingCalls`
+- `diagnostics`, `workspaceDiagnostics`
+- `codeActions` (quickfix, refactor, source.organizeImports, etc.)
+
+---
+
+## Token Caching
+
+Automatic API cost optimization for API key users.
+
+- Caches system instructions and conversation history
+- Reduces tokens processed in subsequent requests
+- Available for API key users (not OAuth)
+- Check savings with `/stats` command
+
+---
+
+## Language / i18n
+
+### UI Language
+
+```bash
+/language ui zh-CN    # Chinese
+/language ui en-US    # English
+/language ui ja-JP    # Japanese
+```
+
+Auto-detection priority: `QWEN_CODE_LANG` env -> `LANG` env -> System locale -> English
+
+### LLM Output Language
+
+```bash
+/language output Chinese
+/language output English
+```
+
+Stored at `~/.qwen/output-language.md`. Restart required after change.
+
+### Custom Language Packs
+
+Create files in `~/.qwen/locales/`:
+
+```javascript
+// ~/.qwen/locales/es.js
+export default {
+  Hello: 'Hola',
+  Settings: 'Configuracion',
+};
+```