(gemini, codex, agy) | API Key Models
(glm, kimi) | -|:--------|:----------------------------------------|:------------------------------| -| **Setup** | Browser auth on first run | Edit settings.json, add API key | -| **Token Storage** | `~/.ccs/cliproxy/auth/
- -### Configuration (Auto-created) - -**CCS automatically creates configuration during installation** (via npm postinstall script). - -**~/.ccs/config.json**: -```json -{ - "profiles": { - "glm": "~/.ccs/glm.settings.json", - "glmt": "~/.ccs/glmt.settings.json", - "kimi": "~/.ccs/kimi.settings.json", - "default": "~/.claude/settings.json" - } -} -``` - -
-
-
-
- -If Claude CLI is installed in a non-standard location (D drive, custom directory), set `CCS_CLAUDE_PATH`: - -```bash -export CCS_CLAUDE_PATH="/path/to/claude" # Unix -$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe" # Windows -``` - -**See also:** [Troubleshooting Guide](./docs/en/troubleshooting.md#claude-cli-in-non-standard-location) for detailed setup instructions. - -
-
-Custom Claude CLI Path
-
-- -If Claude CLI is installed in a non-standard location (D drive, custom directory), set `CCS_CLAUDE_PATH`: - -```bash -export CCS_CLAUDE_PATH="/path/to/claude" # Unix -$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe" # Windows -``` - -**See also:** [Troubleshooting Guide](./docs/en/troubleshooting.md#claude-cli-in-non-standard-location) for detailed setup instructions. - -
-
-
-
- -**Windows users**: Enable Developer Mode for true symlinks (better performance, instant sync): - -1. Open **Settings** → **Privacy & Security** → **For developers** -2. Enable **Developer Mode** -3. Reinstall CCS: `npm install -g @kaitranntt/ccs` - -**Warning:** Without Developer Mode, CCS automatically falls back to copying directories (works but no instant sync across profiles). - -
-
-Windows Symlink Support (Developer Mode)
-
-- -**Windows users**: Enable Developer Mode for true symlinks (better performance, instant sync): - -1. Open **Settings** → **Privacy & Security** → **For developers** -2. Enable **Developer Mode** -3. Reinstall CCS: `npm install -g @kaitranntt/ccs` - -**Warning:** Without Developer Mode, CCS automatically falls back to copying directories (works but no instant sync across profiles). - -
- -## Custom API Profiles (Bring Your Own) - -CCS isn't limited to the built-in providers. **Connect any Anthropic-compatible API** by creating custom profiles: - -```bash -# 1. Create a settings file for your API -cat > ~/.ccs/my-api.settings.json << 'EOF' -{ - "env": { - "ANTHROPIC_BASE_URL": "https://your-api.example.com/v1", - "ANTHROPIC_AUTH_TOKEN": "your-api-key", - "ANTHROPIC_MODEL": "your-model-name" - } -} -EOF - -# 2. Register it in config.json -# Add to ~/.ccs/config.json under "profiles": -# "my-api": "~/.ccs/my-api.settings.json" - -# 3. Use it! -ccs my-api "Your prompt here" -``` - -**Works with any API that implements the Anthropic messages format:** -- Self-hosted LLMs (Ollama, LM Studio, vLLM with Anthropic adapter) -- Alternative AI providers with Anthropic-compatible endpoints -- Enterprise API gateways and proxies -- Custom model deployments - -> [!TIP] -> **Already have API keys for GLM or Kimi?** CCS includes pre-configured profiles for these. Just add your API key to `~/.ccs/glm.settings.json` or `~/.ccs/kimi.settings.json`. - -
- -### Your First Switch - -> [!IMPORTANT] -> **Before using alternative models, update API keys in settings files:** -> -> - **GLM**: Edit `~/.ccs/glm.settings.json` and add your Z.AI Coding Plan API Key -> - **GLMT**: Edit `~/.ccs/glmt.settings.json` and add your Z.AI Coding Plan API Key -> - **Kimi**: Edit `~/.ccs/kimi.settings.json` and add your Kimi API key - -
- -**Parallel Workflow: Planning + Execution** - -```bash -# Terminal 1 - Planning (Claude Sonnet) -ccs "Plan a REST API with authentication and rate limiting" - -# Terminal 2 - Execution (GLM, cost-optimized) -ccs glm "Implement the user authentication endpoints from the plan" -``` - -
-
- -```bash -# Kimi - Stable thinking support -ccs kimi "Design a caching strategy with trade-off analysis" - -# GLMT - Experimental (see full disclaimer below) -ccs glmt "Debug complex algorithm with reasoning steps" -``` - -**Note:** GLMT is experimental and unstable. See [GLM with Thinking (GLMT)](#glm-with-thinking-glmt) section below for full details. - -
-
-Thinking Models (Kimi & GLMT)
- -- -```bash -# Kimi - Stable thinking support -ccs kimi "Design a caching strategy with trade-off analysis" - -# GLMT - Experimental (see full disclaimer below) -ccs glmt "Debug complex algorithm with reasoning steps" -``` - -**Note:** GLMT is experimental and unstable. See [GLM with Thinking (GLMT)](#glm-with-thinking-glmt) section below for full details. - -
- -## The Daily Developer Pain Point - -
-
-### **STOP Switching. START Orchestrating.**
-
-**Session limits shouldn't kill your flow state.**
-
-
-You're deep in implementation. Context loaded. Solution crystallizing.-Then: 🔴 _"You've reached your usage limit."_ - -**Momentum gone. Context lost. Productivity crater.** - -## **The Solution: Parallel Workflows** - -
-
- -```mermaid -graph LR - A[2pm: Building features
In the zone] --> B[3pm: Usage limit hit
BLOCKED] - B --> C[3:05pm: Stop work
Edit settings.json] - C --> D[3:15pm: Switch accounts
Context lost] - D --> E[3:30pm: Restart
Trying to focus] - E --> F[4pm: Finally productive
Back in flow] - - style A fill:#d4edda,stroke:#333,color:#000 - style B fill:#f8d7da,stroke:#333,color:#000 - style C fill:#fff3cd,stroke:#333,color:#000 - style D fill:#f8d7da,stroke:#333,color:#000 - style E fill:#fff3cd,stroke:#333,color:#000 - style F fill:#d4edda,stroke:#333,color:#000 -``` - -**Result:** 1 hour lost, momentum destroyed, frustration builds - -
-
-❌ OLD WAY: Switch When You Hit Limits (Reactive)
- -- -```mermaid -graph LR - A[2pm: Building features
In the zone] --> B[3pm: Usage limit hit
BLOCKED] - B --> C[3:05pm: Stop work
Edit settings.json] - C --> D[3:15pm: Switch accounts
Context lost] - D --> E[3:30pm: Restart
Trying to focus] - E --> F[4pm: Finally productive
Back in flow] - - style A fill:#d4edda,stroke:#333,color:#000 - style B fill:#f8d7da,stroke:#333,color:#000 - style C fill:#fff3cd,stroke:#333,color:#000 - style D fill:#f8d7da,stroke:#333,color:#000 - style E fill:#fff3cd,stroke:#333,color:#000 - style F fill:#d4edda,stroke:#333,color:#000 -``` - -**Result:** 1 hour lost, momentum destroyed, frustration builds - -
-
- -```mermaid -graph LR - A[2pm: Start work] --> B[Terminal 1: Claude Pro
Strategic planning] - A --> C[Terminal 2: GLM
Code execution] - B --> D[3pm: Still shipping
No interruptions] - C --> D - D --> E[4pm: Flow state
Productivity peak] - E --> F[5pm: Features shipped
Context maintained] - - style A fill:#e7f3ff,stroke:#333,color:#000 - style B fill:#cfe2ff,stroke:#333,color:#000 - style C fill:#cfe2ff,stroke:#333,color:#000 - style D fill:#d4edda,stroke:#333,color:#000 - style E fill:#d4edda,stroke:#333,color:#000 - style F fill:#d4edda,stroke:#333,color:#000 -``` - -**Result:** Zero downtime, continuous productivity, less frustration - -### 💰 **The Value Proposition:** -- **Setup:** Your existing Claude Pro + GLM Lite (cost-effective add-on) -- **Value:** Save 1 hour/day × 20 workdays = 20 hours/month recovered -- **ROI:** Your development time is worth more than the setup cost -- **Reality:** Shipping faster than the overhead - -
-
-## Choose Your Path
-
-✨ NEW WAY: Run Parallel From Start (Proactive) - RECOMMENDED
- -- -```mermaid -graph LR - A[2pm: Start work] --> B[Terminal 1: Claude Pro
Strategic planning] - A --> C[Terminal 2: GLM
Code execution] - B --> D[3pm: Still shipping
No interruptions] - C --> D - D --> E[4pm: Flow state
Productivity peak] - E --> F[5pm: Features shipped
Context maintained] - - style A fill:#e7f3ff,stroke:#333,color:#000 - style B fill:#cfe2ff,stroke:#333,color:#000 - style C fill:#cfe2ff,stroke:#333,color:#000 - style D fill:#d4edda,stroke:#333,color:#000 - style E fill:#d4edda,stroke:#333,color:#000 - style F fill:#d4edda,stroke:#333,color:#000 -``` - -**Result:** Zero downtime, continuous productivity, less frustration - -### 💰 **The Value Proposition:** -- **Setup:** Your existing Claude Pro + GLM Lite (cost-effective add-on) -- **Value:** Save 1 hour/day × 20 workdays = 20 hours/month recovered -- **ROI:** Your development time is worth more than the setup cost -- **Reality:** Shipping faster than the overhead - -
-
-
-Budget-Focused: GLM Only
- -- **Best for:** Cost-conscious development, basic code generation -- **Usage:** Just use `ccs glm` directly for cost-effective AI assistance -- **Reality:** No Claude access, but capable for many coding tasks -- **Setup:** GLM API key only, very affordable - -
-
-
-✨ Recommended for Daily Development: 1 Claude Pro + 1 GLM Lite
- -- **Best for:** Daily code delivery, serious development work -- **Usage:** `ccs` for planning + `ccs glm` for execution (parallel workflow) -- **Reality:** Perfect balance of capability and cost for most developers -- **Value:** Never hit session limits, continuous productivity - -
-
-
-Power User: Multiple Claude Pro + GLM Pro
- -- **Best for:** Heavy workloads, concurrent projects, solo dev -- **Unlocks:** Never drain session or weekly limits -- **Workflow:** 3+ terminals running specialized tasks simultaneously - -
-
-
----
-
-## Why CCS Instead of Manual Switching?
-
-Privacy-Focused: Work/Personal Isolation
- -- **When needed:** Strict separation of work and personal AI contexts -- **Setup:** `ccs auth create work` + `ccs auth create personal` -- **Note:** Advanced feature - most users don't need this - -
-
-**CCS isn't about "switching when you hit limits at 3pm."**
-
-## **It's about running in parallel from the start.**
-
-
-
-### The Core Difference
-
-| **Manual Switching** | **CCS Orchestration** |
-|:---|:---|
-| 🔴 Hit limits → Stop work → Edit config files → Restart | ✅ Multiple terminals running different models from the start |
-| 😰 Context loss and flow state interruption | 😌 Continuous productivity with preserved context |
-| 📝 Sequential task handling | ⚡ Parallel workflows (planning + execution simultaneously) |
-| 🛠️ Reactive problem solving when blocked | 🎯 Proactive workflow design prevents blocks |
-
-### What CCS Gives You
-
-- **Unlimited Profiles**: Built-in OAuth, API keys, or bring your own Anthropic-compatible endpoints
-- **Zero Context Switching**: Keep your flow state without interruption
-- **Parallel Productivity**: Strategic planning in one terminal, code execution in another
-- **Instant Account Management**: One command switches, no config file editing
-- **Work-Life Separation**: Isolate contexts without logging out
-- **Cross-Platform Consistency**: Same smooth experience on macOS, Linux, Windows
-
-**Manual context switching breaks workflow. CCS orchestrates seamlessly.**
-
-- -## Architecture - -CCS is a **universal profile manager** that resolves profiles in priority order: -1. **CLIProxy profiles** (gemini, codex, agy) - Built-in OAuth providers, zero config -2. **CLIProxy variants** (user-defined) - Custom model configurations for CLIProxy providers -3. **Settings-based profiles** (glm, kimi, **your-custom-api**) - Any Anthropic-compatible API -4. **Account-based profiles** (work, personal) - Isolated Claude instances -5. **Default** - Claude CLI with subscription - -> [!NOTE] -> **Extensibility is core to CCS.** While we provide built-in profiles for common providers, you can connect **any API** that speaks the Anthropic messages format. - -### Profile Types - -**CLIProxy**: gemini, codex, agy (v5.0+) -- OAuth-based authentication via [CLIProxyAPI](https://github.com/router-for-me/CLIProxyAPI) -- Browser auth on first run, tokens cached in `~/.ccs/cliproxy/auth/
- -## Web Dashboard - -CCS includes a modern React 19 dashboard for visual profile management and real-time monitoring: - -```bash -# Start the web dashboard (auto-detects available port) ccs config - -# Or specify a port -ccs config --port 3000 - -# Access at http://localhost:PORT (shown in output) +# Opens http://localhost:3000 ``` -**Dashboard Features**: -- 🎨 **Modern UI**: Built with React 19, TypeScript, and shadcn/ui -- 📊 **Real-time Updates**: WebSocket integration for live status -- ⚙️ **Profile Management**: Visual configuration of all profiles -- 🔍 **Health Monitoring**: System diagnostics and metrics -- 🌙 **Dark Mode**: Eye-friendly theme switching -- 📱 **Responsive**: Works on desktop and mobile +### 3. Configure Your Accounts -### Dashboard Screenshots +The dashboard provides visual management for all account types: -*(Add screenshots here when available)* +- **Claude Accounts**: Create isolated instances (work, personal, client) +- **OAuth Providers**: One-click auth for Gemini, Codex, Antigravity +- **API Profiles**: Configure GLM, Kimi with your keys +- **Health Monitor**: Real-time status across all profiles -## Usage Examples +**Analytics (Light/Dark Theme)** -### Basic Switching + -```bash -ccs # Claude subscription (default) -ccs glm # GLM (cost-optimized) -ccs kimi # Kimi (with thinking support) -``` + -### Multi-Account Setup +**API Profiles & OAuth Providers** -```bash -# Create accounts -ccs auth create work -ccs auth create personal -``` + -**Run concurrently in separate terminals:** - -```bash -# Terminal 1 - Work -ccs work "implement feature" - -# Terminal 2 - Personal (concurrent) -ccs personal "review code" -``` - -### Help & Version - -```bash -ccs --version # Show version -ccs --help # Show all commands and options -``` +
-## AI-Powered Delegation +## Built-in Providers + +| Provider | Auth Type | Command | Best For | +|----------|-----------|---------|----------| +| **Claude** | Subscription | `ccs` | Default, strategic planning | +| **Gemini** | OAuth | `ccs gemini` | Zero-config, fast iteration | +| **Codex** | OAuth | `ccs codex` | Code generation | +| **Antigravity** | OAuth | `ccs agy` | Alternative routing | +| **GLM** | API Key | `ccs glm` | Cost-optimized execution | +| **Kimi** | API Key | `ccs kimi` | Long-context, thinking mode | + +> **OAuth providers** authenticate via browser on first run. Tokens are cached in `~/.ccs/cliproxy/auth/`. > [!TIP] -> **New in v4.0**: Delegate tasks to cost-optimized models (GLM, Kimi) directly from your main Claude session. Save 81% on simple tasks with real-time visibility. - -### What is Delegation? - -CCS Delegation lets you **send tasks to alternative models** (`glm`, `kimi`) **from your main Claude session** using the `-p` flag or intelligent slash commands (`/ccs`, `/ccs:continue`). - -**Why use it?** -- **Token efficiency**: Simple tasks cost 81% less on GLM vs main Claude session -- **Context preservation**: Main session stays clean, no pollution from mechanical tasks -- **Real-time visibility**: See tool usage as tasks execute (`[Tool] Write: index.html`) -- **Multi-turn support**: Resume sessions with `:continue` for iterative work - -### Quick Examples - -**Direct CLI:** -```bash -# Delegate simple task to GLM (cost-optimized) -ccs glm -p "add tests for UserService" - -# Delegate long-context task to Kimi -ccs kimi -p "analyze all files in src/ and document architecture" - -# Continue previous session -ccs glm:continue -p "run the tests and fix any failures" -``` - -**Via Slash Commands** (inside Claude sessions): -```bash -# In your main Claude session: -/ccs "refactor auth.js to use async/await" # Auto-selects best profile -/ccs --glm "fix typo in documentation" # Forces GLM profile -/ccs --kimi "analyze entire architecture" # Forces Kimi profile -/ccs:continue "also update the README examples" # Continues last session -``` - -**Via Natural Language** (Claude auto-delegates): -```bash -# Claude detects delegation patterns and auto-executes: -"Use ccs glm to add tests for all *.service.js files" -"Delegate to kimi: analyze project structure" -``` - -### Real-Time Output - -See exactly what's happening as tasks execute: - -``` -$ ccs glm -p "/cook create a landing page" -[i] Delegating to GLM-4.6... -[Tool] Write: /home/user/project/index.html -[Tool] Write: /home/user/project/styles.css -[Tool] Write: /home/user/project/script.js -[Tool] Edit: /home/user/project/styles.css -[i] Execution completed in 45.2s - -╔══════════════════════════════════════════════════════╗ -║ Working Directory: /home/user/project ║ -║ Model: GLM-4.6 ║ -║ Duration: 45.2s ║ -║ Exit Code: 0 ║ -║ Session ID: 3a4f8c21 ║ -║ Total Cost: $0.0015 ║ -║ Turns: 3 ║ -╚══════════════════════════════════════════════════════╝ -``` - -### Advanced Features - -**Slash Command Support:** -Delegation preserves custom slash commands in prompts: -```bash -ccs glm -p "/cook create responsive landing page" -# Executes /cook command in delegated GLM session -``` - -**Signal Handling:** -Ctrl+C or Esc properly kills delegated processes (no orphans): -```bash -# Hit Ctrl+C during delegation -[!] Parent process terminating, killing delegated session... -``` - -**Time-Based Limits:** -10-minute default timeout with graceful termination (supports `:continue`): -```bash -ccs glm -p "complex task" # Auto-terminates after 10min if needed -ccs glm:continue -p "pick up where we left off" -``` - -### Cost Savings Example - -**Traditional (Main Session):** -``` -Context load: 2000 tokens -Discussion: 1500 tokens -Code gen: 4500 tokens -───────────────────────── -Total: 8000 tokens → $0.032 -``` - -**Delegation (GLM):** -``` -3x tasks via GLM: 1500 tokens → $0.0045 -───────────────────────────────────────── -Savings: $0.0275 (86% reduction) -``` - -### Documentation - -- **Workflow Diagrams**: See [docs/ccs-delegation-diagrams.md](docs/ccs-delegation-diagrams.md) for visual architecture -- **Skill Reference**: `.claude/skills/ccs-delegation/` for AI decision framework (replaces deprecated agents) -- **Agent Docs**: `.claude/agents/ccs-delegator.md` was deprecated in v4.3.2, functionality moved to ccs-delegation skill +> **Need more?** CCS supports **any Anthropic-compatible API**. Create custom profiles for self-hosted LLMs, enterprise gateways, or alternative providers. See [API Profiles documentation](https://docs.ccs.kaitran.ca/providers/api-profiles).
-## GLM with Thinking (GLMT) +## Usage -> [!CAUTION] -> ### NOT PRODUCTION READY - EXPERIMENTAL FEATURE -> -> **GLMT is experimental and requires extensive debugging**: -> - Streaming and tool support still under active development -> - May experience unexpected errors, timeouts, or incomplete responses -> - Requires frequent debugging and manual intervention -> - **Not recommended for critical workflows or production use** -> -> **Alternative for GLM Thinking**: Consider going through the **CCR hustle** with the **Transformer of Bedolla** ([ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/)) for a more stable implementation. - -> [!IMPORTANT] -> GLMT requires npm installation (`npm install -g @kaitranntt/ccs`). Not available in native shell versions (requires Node.js HTTP server). - -
- -> [!NOTE] -> ### Acknowledgments: The Foundation That Made GLMT Possible -> -> **CCS's GLMT implementation owes its existence to the groundbreaking work of [@Bedolla](https://github.com/Bedolla)**, who created [ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/) - the **first integration** to bridge [Claude Code Router (CCR)](https://github.com/musistudio/claude-code-router) with Z.AI's reasoning capabilities. -> -> **Why this matters**: Before ZaiTransformer, no one had successfully integrated Z.AI's thinking mode with Claude Code's workflow. Bedolla's work wasn't just helpful - it was **foundational**. His implementation of: -> -> - **Request/response transformation architecture** - The conceptual blueprint for how to bridge Anthropic and OpenAI formats -> - **Thinking mode control mechanisms** - The patterns for managing reasoning_content delivery -> - **Embedded proxy design** - The architecture that CCS's GLMT proxy is built upon -> -> These contributions directly inspired and enabled GLMT's design. **Without ZaiTransformer's pioneering work, GLMT wouldn't exist in its current form**. The technical patterns, transformation logic, and proxy architecture implemented in CCS are a direct evolution of the concepts Bedolla first proved viable. -> -> **Recognition**: If you benefit from GLMT's thinking capabilities, you're benefiting from Bedolla's vision and engineering. Please consider starring [ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/) to support pioneering work in the Claude Code ecosystem. - -
- -
-
-
-
- -
-
-GLM vs GLMT Comparison
-
-- -
-
-| Feature | GLM (`ccs glm`) | GLMT (`ccs glmt`) |
-|:--------|:----------------|:------------------|
-| **Endpoint** | Anthropic-compatible | OpenAI-compatible |
-| **Thinking** | No | Experimental (`reasoning_content`) |
-| **Tool Support** | Basic | **Unstable (v3.5+)** |
-| **MCP Tools** | Limited | **Buggy (v3.5+)** |
-| **Streaming** | Stable | **Experimental (v3.4+)** |
-| **TTFB** | <500ms | <500ms (sometimes), 2-10s+ (often) |
-| **Use Case** | Reliable work | **Debugging experiments only** |
-
-
-
-
-
-
-
- -**GLMT attempts MCP tools and function calling:** - -- **Bidirectional Transformation**: Anthropic tools ↔ OpenAI format (unstable) -- **MCP Integration**: MCP tools sometimes execute (often output XML garbage) -- **Streaming Tool Calls**: Real-time tool calls (when not crashing) -- **Backward Compatible**: May break existing thinking support -- **Configuration Required**: Frequent manual debugging needed - -
-
-Tool Support (v3.5) - EXPERIMENTAL
-
-- -**GLMT attempts MCP tools and function calling:** - -- **Bidirectional Transformation**: Anthropic tools ↔ OpenAI format (unstable) -- **MCP Integration**: MCP tools sometimes execute (often output XML garbage) -- **Streaming Tool Calls**: Real-time tool calls (when not crashing) -- **Backward Compatible**: May break existing thinking support -- **Configuration Required**: Frequent manual debugging needed - -
-
-
-
- -**GLMT attempts real-time streaming** with incremental reasoning content delivery: - -- **Default**: Streaming enabled (TTFB <500ms when it works) -- **Auto-fallback**: Frequently switches to buffered mode due to errors -- **Thinking parameter**: Claude CLI `thinking` parameter sometimes works - - May ignore `thinking.type` and `budget_tokens` - - Precedence: CLI parameter > message tags > default (when not broken) - -**Status**: Z.AI (tested, tool calls frequently break, requires constant debugging) - -
-
-Streaming Support (v3.4) - OFTEN FAILS
-
-- -**GLMT attempts real-time streaming** with incremental reasoning content delivery: - -- **Default**: Streaming enabled (TTFB <500ms when it works) -- **Auto-fallback**: Frequently switches to buffered mode due to errors -- **Thinking parameter**: Claude CLI `thinking` parameter sometimes works - - May ignore `thinking.type` and `budget_tokens` - - Precedence: CLI parameter > message tags > default (when not broken) - -**Status**: Z.AI (tested, tool calls frequently break, requires constant debugging) - -
-
-
-
- -1. CCS spawns embedded HTTP proxy on localhost (if not crashing) -2. Proxy attempts to convert Anthropic format → OpenAI format (often fails) -3. Tries to transform Anthropic tools → OpenAI function calling format (buggy) -4. Forwards to Z.AI with reasoning parameters and tools (when not timing out) -5. Attempts to convert `reasoning_content` → thinking blocks (partial or broken) -6. Attempts to convert OpenAI `tool_calls` → Anthropic `tool_use` blocks (XML garbage common) -7. Thinking and tool calls sometimes appear in Claude Code UI (when not broken) - -
-
-How It Works (When It Works)
-
-- -1. CCS spawns embedded HTTP proxy on localhost (if not crashing) -2. Proxy attempts to convert Anthropic format → OpenAI format (often fails) -3. Tries to transform Anthropic tools → OpenAI function calling format (buggy) -4. Forwards to Z.AI with reasoning parameters and tools (when not timing out) -5. Attempts to convert `reasoning_content` → thinking blocks (partial or broken) -6. Attempts to convert OpenAI `tool_calls` → Anthropic `tool_use` blocks (XML garbage common) -7. Thinking and tool calls sometimes appear in Claude Code UI (when not broken) - -
-
-
-
- -**Control Tags**: -- `` - Enable/disable reasoning blocks (default: On)
-- `` - Control reasoning depth (deprecated - Z.AI only supports binary thinking)
-
-**Thinking Keywords** (inconsistent activation):
-- `think` - Sometimes enables reasoning (low effort)
-- `think hard` - Sometimes enables reasoning (medium effort)
-- `think harder` - Sometimes enables reasoning (high effort)
-- `ultrathink` - Attempts maximum reasoning depth (often breaks)
-
-
-
-Control Tags & Keywords
-
-- -**Control Tags**: -- `
-
-
-
- -**GLMT features** (all experimental): -- Forced English output enforcement (sometimes works) -- Random thinking mode activation (unpredictable) -- Attempted streaming with frequent fallback to buffered mode - -**General**: -- `CCS_DEBUG_LOG=1` - Enable debug file logging -- `CCS_CLAUDE_PATH=/path/to/claude` - Custom Claude CLI path - -
-
-Environment Variables
-
-- -**GLMT features** (all experimental): -- Forced English output enforcement (sometimes works) -- Random thinking mode activation (unpredictable) -- Attempted streaming with frequent fallback to buffered mode - -**General**: -- `CCS_DEBUG_LOG=1` - Enable debug file logging -- `CCS_CLAUDE_PATH=/path/to/claude` - Custom Claude CLI path - -
-
-
-
+### Basic Commands ```bash -# Edit GLMT settings -nano ~/.ccs/glmt.settings.json +ccs # Default Claude session +ccs agy # Antigravity (OAuth) +ccs gemini # Gemini (OAuth) +ccs glm # GLM (API key) ``` -Set Z.AI API key (requires coding plan): +### Parallel Workflows -```json -{ - "env": { - "ANTHROPIC_AUTH_TOKEN": "your-z-ai-api-key" - } -} -``` +Run multiple terminals with different providers: -
-
-API Key Setup
-
-+### Basic Commands ```bash -# Edit GLMT settings -nano ~/.ccs/glmt.settings.json +ccs # Default Claude session +ccs agy # Antigravity (OAuth) +ccs gemini # Gemini (OAuth) +ccs glm # GLM (API key) ``` -Set Z.AI API key (requires coding plan): +### Parallel Workflows -```json -{ - "env": { - "ANTHROPIC_AUTH_TOKEN": "your-z-ai-api-key" - } -} -``` +Run multiple terminals with different providers: -
-
-
-
- -**v3.4 Protection Limits**: - -| Limit | Value | Purpose | -|:------|:------|:--------| -| **SSE buffer** | 1MB max per event | Prevent buffer overflow | -| **Content buffer** | 10MB max per block | Limit thinking/text blocks | -| **Content blocks** | 100 max per message | Prevent DoS attacks | -| **Request timeout** | 120s | Both streaming and buffered | - -
-
-Security Limits (DoS Protection)
-
-- -**v3.4 Protection Limits**: - -| Limit | Value | Purpose | -|:------|:------|:--------| -| **SSE buffer** | 1MB max per event | Prevent buffer overflow | -| **Content buffer** | 10MB max per block | Limit thinking/text blocks | -| **Content blocks** | 100 max per message | Prevent DoS attacks | -| **Request timeout** | 120s | Both streaming and buffered | - -
-
-
-
- -**Enable verbose logging**: ```bash -ccs glmt --verbose "your prompt" +# Terminal 1: Planning (Claude Pro) +ccs work "design the authentication system" + +# Terminal 2: Execution (GLM - cost optimized) +ccs glm "implement the user service from the plan" + +# Terminal 3: Review (Gemini) +ccs gemini "review the implementation for security issues" ``` -**Enable debug file logging**: +### Multi-Account Claude + +Create isolated Claude instances for work/personal separation: + ```bash -export CCS_DEBUG_LOG=1 -ccs glmt --verbose "your prompt" -# Logs: ~/.ccs/logs/ +ccs auth create work + +# Run concurrently in separate terminals +ccs work "implement feature" # Terminal 1 +ccs "review code" # Terminal 2 (personal account) ``` -**GLMT debugging**: -```bash -# Verbose logging shows streaming status and reasoning details -ccs glmt --verbose "test" -``` - -**Check reasoning content**: -```bash -cat ~/.ccs/logs/*response-openai.json | jq '.choices[0].message.reasoning_content' -``` - -**Troubleshooting**: -- **If absent**: Z.AI API issue (verify key, account status) -- **If present**: Transformation issue (check `response-anthropic.json`) - -
-
Debugging
-
-- -**Enable verbose logging**: ```bash -ccs glmt --verbose "your prompt" +# Terminal 1: Planning (Claude Pro) +ccs work "design the authentication system" + +# Terminal 2: Execution (GLM - cost optimized) +ccs glm "implement the user service from the plan" + +# Terminal 3: Review (Gemini) +ccs gemini "review the implementation for security issues" ``` -**Enable debug file logging**: +### Multi-Account Claude + +Create isolated Claude instances for work/personal separation: + ```bash -export CCS_DEBUG_LOG=1 -ccs glmt --verbose "your prompt" -# Logs: ~/.ccs/logs/ +ccs auth create work + +# Run concurrently in separate terminals +ccs work "implement feature" # Terminal 1 +ccs "review code" # Terminal 2 (personal account) ``` -**GLMT debugging**: -```bash -# Verbose logging shows streaming status and reasoning details -ccs glmt --verbose "test" -``` - -**Check reasoning content**: -```bash -cat ~/.ccs/logs/*response-openai.json | jq '.choices[0].message.reasoning_content' -``` - -**Troubleshooting**: -- **If absent**: Z.AI API issue (verify key, account status) -- **If present**: Transformation issue (check `response-anthropic.json`) - -
## Maintenance ### Health Check -Run diagnostics to verify your CCS installation: - ```bash ccs doctor ``` -**Checks performed**: -- ✓ Claude CLI availability -- ✓ Configuration files (config.json, profiles) -- ✓ CCS symlinks to ~/.claude/ -- ✓ Delegation system -- ✓ File permissions +Verifies: Claude CLI, config files, symlinks, permissions. -**Output**: -``` -[?] Checking Claude CLI... [OK] -[?] Checking ~/.ccs/ directory... [OK] -[?] Checking config.json... [OK] -[?] Checking CCS symlinks... [OK] -... -Status: Installation healthy -``` - -### Updating CCS +### Update ```bash -# Update to latest stable version -ccs update - -# Force reinstall (fix corrupted installation) -ccs update --force - -# Install beta/dev channel (unstable, for testing) -ccs update --beta - -# Force reinstall from dev channel -ccs update --force --beta +ccs update # Update to latest +ccs update --force # Force reinstall +ccs update --beta # Install dev channel ``` -**Note:** `--beta` installs from the `@dev` npm tag, which contains unreleased features. Not recommended for production use. - -### Update CCS Items - -If you modify CCS items or need to re-install symlinks: +### Sync Shared Items ```bash ccs sync ``` -**What it does**: -- Re-creates selective symlinks from `~/.ccs/.claude/` to `~/.claude/` -- Backs up existing files before replacing -- Safe to run multiple times (idempotent) - -**When to use**: -- After manual modifications to ~/.claude/ -- If `ccs doctor` reports symlink issues -- After upgrading CCS to a new version +Re-creates symlinks for shared commands, skills, and settings.
-### Troubleshooting +## Configuration -#### Broken Installation +CCS auto-creates config on install. Dashboard is the recommended way to manage settings. -If CCS is corrupted or behaving unexpectedly: +**Config location**: `~/.ccs/config.yaml` + +
+
-```bash
-npm install -g @kaitranntt/ccs@latest --force
-```
+Custom Claude CLI path
+ +If Claude CLI is installed in a non-standard location: ```bash -# Force reinstall to fix -ccs update --force +export CCS_CLAUDE_PATH="/path/to/claude" # Unix +$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe" # Windows ``` -If that doesn't work, try manual reinstall: +
+
-- **Update fails**: Check network connection and try `ccs update --force`
-- **Beta not working**: Use `ccs update` to return to stable version
-- **Direct install error**: Beta channel requires npm installation
+Windows symlink support
-#### Beta Testing Issues +Enable Developer Mode for true symlinks: -If you're on the dev channel and experiencing issues: +1. **Settings** → **Privacy & Security** → **For developers** +2. Enable **Developer Mode** +3. Reinstall: `npm install -g @kaitranntt/ccs` -```bash -# Return to stable channel -ccs update -``` +Without Developer Mode, CCS falls back to copying directories. -#### Common Issues +-For detailed troubleshooting, see [Troubleshooting Guide](./docs/en/troubleshooting.md). +## Documentation + +| Topic | Link | +|-------|------| +| Installation | [docs.ccs.kaitran.ca/getting-started/installation](https://docs.ccs.kaitran.ca/getting-started/installation) | +| Configuration | [docs.ccs.kaitran.ca/getting-started/configuration](https://docs.ccs.kaitran.ca/getting-started/configuration) | +| OAuth Providers | [docs.ccs.kaitran.ca/providers/oauth-providers](https://docs.ccs.kaitran.ca/providers/oauth-providers) | +| Multi-Account Claude | [docs.ccs.kaitran.ca/providers/claude-accounts](https://docs.ccs.kaitran.ca/providers/claude-accounts) | +| API Profiles | [docs.ccs.kaitran.ca/providers/api-profiles](https://docs.ccs.kaitran.ca/providers/api-profiles) | +| CLI Reference | [docs.ccs.kaitran.ca/reference/cli-commands](https://docs.ccs.kaitran.ca/reference/cli-commands) | +| Architecture | [docs.ccs.kaitran.ca/reference/architecture](https://docs.ccs.kaitran.ca/reference/architecture) | +| Troubleshooting | [docs.ccs.kaitran.ca/reference/troubleshooting](https://docs.ccs.kaitran.ca/reference/troubleshooting) |
## Uninstall -### npm (Recommended) - ```bash npm uninstall -g @kaitranntt/ccs ```
-
+
-Alternative Package Managers
- -+
Alternative package managers
```bash -# yarn yarn global remove @kaitranntt/ccs - -# pnpm pnpm remove -g @kaitranntt/ccs - -# bun bun remove -g @kaitranntt/ccs ```
-
-```bash -# macOS / Linux -curl -fsSL ccs.kaitran.ca/uninstall | bash - -# Windows PowerShell -irm ccs.kaitran.ca/uninstall | iex -``` - -
-
-Legacy Uninstallers (for native installs)
--```bash -# macOS / Linux -curl -fsSL ccs.kaitran.ca/uninstall | bash - -# Windows PowerShell -irm ccs.kaitran.ca/uninstall | iex -``` - -
- -## 🎯 Philosophy +## Philosophy - **YAGNI**: No features "just in case" -- **KISS**: Simple bash, no complexity +- **KISS**: Simple, focused implementation - **DRY**: One source of truth (config) -## 📖 Documentation +
-**Complete documentation in [docs/](./docs/)**: -- [Installation Guide](./docs/en/installation.md) -- [Configuration](./docs/en/configuration.md) -- [Usage Examples](./docs/en/usage.md) -- [System Architecture](./docs/system-architecture.md) -- [GLMT Control Mechanisms](./docs/glmt-controls.md) -- [Troubleshooting](./docs/en/troubleshooting.md) -- [Contributing](./CONTRIBUTING.md) +## Contributing -## 🤝 Contributing - -We welcome contributions! Please see our [Contributing Guide](./CONTRIBUTING.md) for details. - - -## Star History - -
-
-
-
-
+See [CONTRIBUTING.md](./CONTRIBUTING.md).
+## License -CCS is licensed under the [MIT License](LICENSE). +MIT License - see [LICENSE](LICENSE).
-**Made with ❤️ for developers who hit rate limits too often**
+---
-[⭐ Star this repo](https://github.com/kaitranntt/ccs) | [🐛 Report issues](https://github.com/kaitranntt/ccs/issues) | [📖 Read docs](./docs/en/)
+**[ccs.kaitran.ca](https://ccs.kaitran.ca)** | [Report Issues](https://github.com/kaitranntt/ccs/issues) | [Star on GitHub](https://github.com/kaitranntt/ccs)
diff --git a/VERSION b/VERSION
index ad95545a..1f6ad15d 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-5.17.0
+5.17.0-dev.7
diff --git a/docs/assets/ccs-logo-medium.png b/assets/ccs-logo-medium.png
similarity index 100%
rename from docs/assets/ccs-logo-medium.png
rename to assets/ccs-logo-medium.png
diff --git a/docs/assets/ccs-logo-zoomed-web.png b/assets/ccs-logo-zoomed-web.png
similarity index 100%
rename from docs/assets/ccs-logo-zoomed-web.png
rename to assets/ccs-logo-zoomed-web.png
diff --git a/assets/screenshots/analytics-light.png b/assets/screenshots/analytics-light.png
new file mode 100644
index 00000000..3ab3f969
Binary files /dev/null and b/assets/screenshots/analytics-light.png differ
diff --git a/assets/screenshots/analytics.png b/assets/screenshots/analytics.png
new file mode 100644
index 00000000..34bd1563
Binary files /dev/null and b/assets/screenshots/analytics.png differ
diff --git a/assets/screenshots/api_profiles.png b/assets/screenshots/api_profiles.png
new file mode 100644
index 00000000..b6243bc1
Binary files /dev/null and b/assets/screenshots/api_profiles.png differ
diff --git a/assets/screenshots/cliproxy.png b/assets/screenshots/cliproxy.png
new file mode 100644
index 00000000..a292e6a0
Binary files /dev/null and b/assets/screenshots/cliproxy.png differ
diff --git a/docs/ccs-delegation-diagrams.md b/docs/ccs-delegation-diagrams.md
deleted file mode 100644
index 0cf8df2f..00000000
--- a/docs/ccs-delegation-diagrams.md
+++ /dev/null
@@ -1,492 +0,0 @@
-# CCS Delegation Workflow Diagrams
-
-Visual guide to understanding how CCS delegation works internally.
-
----
-
-## Overview Architecture
-
-```
-┌─────────────────────────────────────────────────────────────┐
-│ CCS Architecture │
-├─────────────────────────────────────────────────────────────┤
-│ │
-│ User Input │
-│ │ │
-│ ├─── ccs glm → Normal Profile Execution │
-│ │ │
-│ └─── ccs glm -p "task" → Delegation Flow ⚡ │
-│ │
-└─────────────────────────────────────────────────────────────┘
-```
-
----
-
-## Execution Flow Comparison
-
-### Normal Execution (Without -p)
-```
-User: ccs glm
- │
- ├─→ bin/ccs.js (main entry)
- │ │
- │ ├─→ Profile Detection: "glm"
- │ │
- │ └─→ execClaude()
- │ │
- │ └─→ spawn("claude", ["--settings", "~/.ccs/glm.settings"])
- │ │
- │ └─→ Claude CLI Interactive Session
- │ │
- │ └─→ Direct User Interaction
-```
-
-### Delegation Execution (With -p)
-```
-User: ccs glm -p "add tests for UserService"
- │
- ├─→ bin/ccs.js (main entry)
- │ │
- │ ├─→ -p Flag Detected! 🎯
- │ │
- │ └─→ DelegationHandler.route(args)
- │ │
- │ ├─→ Parse args
- │ │ ├─ profile: "glm"
- │ │ ├─ prompt: "add tests for UserService"
- │ │ └─ options: { outputFormat: "stream-json", timeout: 600000 }
- │ │
- │ ├─→ Validate profile (DelegationValidator)
- │ │
- │ └─→ HeadlessExecutor.execute("glm", prompt, options)
- │ │
- │ ├─→ spawn("claude", [
- │ │ "-p", prompt,
- │ │ "--settings", "~/.ccs/glm.settings",
- │ │ "--output-format", "stream-json",
- │ │ "--permission-mode", "acceptEdits"
- │ │ ])
- │ │
- │ ├─→ Parse stream-JSON output (jsonl format)
- │ │ {"type":"init","session_id":"abc123"}
- │ │ {"type":"assistant","message":{...}}
- │ │ {"type":"result","total_cost_usd":0.0042,"num_turns":3}
- │ │
- │ ├─→ SessionManager.saveSession()
- │ │ └─→ ~/.ccs/delegation-sessions.json
- │ │
- │ └─→ ResultFormatter.format(result)
- │ │
- │ └─→ ASCII Box Output
- │ ╔════════════════════════╗
- │ ║ Session: abc123 ║
- │ ║ Cost: $0.0042 ║
- │ ║ Turns: 3 ║
- │ ╚════════════════════════╝
-```
-
----
-
-## Continue Command Flow
-
-### Multi-Turn Session Workflow
-```
-┌─────────────────────────────────────────────────────────────┐
-│ Turn 1: Initial Task │
-└─────────────────────────────────────────────────────────────┘
-
-User: ccs glm -p "implement user registration"
- │
- └─→ HeadlessExecutor
- │
- ├─→ Execute with fresh session
- │
- └─→ Save session metadata:
- {
- "profile": "glm",
- "sessionId": "session-001",
- "totalCost": 0.0025,
- "turns": 2,
- "cwd": "/path/to/project",
- "lastUpdated": "2025-11-15T18:00:00Z"
- }
-
-┌─────────────────────────────────────────────────────────────┐
-│ Turn 2: Continue Session │
-└─────────────────────────────────────────────────────────────┘
-
-User: ccs glm:continue -p "add validation tests"
- │
- └─→ DelegationHandler detects ":continue" suffix
- │
- ├─→ Extract base profile: "glm"
- │
- ├─→ SessionManager.getLastSession("glm")
- │ └─→ Returns: { sessionId: "session-001", ... }
- │
- └─→ HeadlessExecutor.execute("glm", prompt, {
- resumeSession: true,
- sessionId: "session-001" ← Resume!
- })
- │
- ├─→ spawn("claude", [
- │ "-p", "add validation tests",
- │ "--resume", "session-001", ← Continue!
- │ "--output-format", "stream-json",
- │ "--verbose",
- │ ...
- │ ])
- │
- └─→ Update session metadata:
- {
- "sessionId": "session-001", ← Same session
- "totalCost": 0.0067, ← Aggregated
- "turns": 5, ← Incremented
- "lastUpdated": "2025-11-15T18:05:00Z"
- }
-
-┌─────────────────────────────────────────────────────────────┐
-│ Turn 3+: Multiple Continues │
-└─────────────────────────────────────────────────────────────┘
-
-User: ccs glm:continue -p "run the tests"
- │
- └─→ Same flow, cost keeps aggregating:
- {
- "totalCost": 0.0089, ← $0.0025 + $0.0042 + $0.0022
- "turns": 7 ← 2 + 3 + 2
- }
-```
-
----
-
-## Session Management Architecture
-
-```
-┌───────────────────────────────────────────────────────────────┐
-│ ~/.ccs/delegation-sessions.json │
-├───────────────────────────────────────────────────────────────┤
-│ { │
-│ "glm": { │
-│ "sessionId": "abc123-def456", │
-│ "totalCost": 0.0067, ← Aggregated across turns │
-│ "turns": 5, ← Total turn count │
-│ "cwd": "/home/user/project", ← Working directory │
-│ "lastUpdated": "2025-11-15T18:05:00Z", │
-│ "expiresAt": "2025-12-15T18:05:00Z" ← 30 days │
-│ }, │
-│ "kimi": { │
-│ "sessionId": "xyz789-uvw012", │
-│ "totalCost": 0.0123, │
-│ "turns": 8, │
-│ "cwd": "/home/user/other-project", │
-│ "lastUpdated": "2025-11-14T10:30:00Z", │
-│ "expiresAt": "2025-12-14T10:30:00Z" │
-│ } │
-│ } │
-└───────────────────────────────────────────────────────────────┘
-
-Operations:
- ├─→ saveSession(profile, metadata) → Write to file
- ├─→ getLastSession(profile) → Read from file
- ├─→ updateSession(profile, updates) → Merge + write
- └─→ cleanupExpired() → Remove old sessions (>30 days)
-```
-
----
-
-## Decision Flow: When to Delegate
-
-```
-User Task Request
- │
- ├─→ Task Analysis (ccs-delegator agent)
- │ │
- │ ├─→ Read ccs-delegation skill
- │ │
- │ └─→ Pattern Matching:
- │ │
- │ ├─→ Match delegation patterns?
- │ │ ├─ "refactor .* to use async/await" ✓
- │ │ ├─ "add tests for .*" ✓
- │ │ ├─ "fix typos in .*" ✓
- │ │ └─ ...
- │ │
- │ ├─→ Match anti-patterns?
- │ │ ├─ "implement .*" ✗
- │ │ ├─ "optimize .*" ✗
- │ │ └─ "design .*" ✗
- │ │
- │ └─→ Check criteria:
- │ ├─ Scope: < 5 files? ✓
- │ ├─ Complexity: Mechanical? ✓
- │ ├─ Ambiguity: Zero decisions? ✓
- │ └─ Context: Patterns exist? ✓
- │
- └─→ Decision:
- │
- ├─→ YES → Delegate
- │ └─→ ccs glm -p "task"
- │
- └─→ NO → Keep in main session
- └─→ Handle directly in conversation
-```
-
----
-
-## Cost Tracking & Token Optimization
-
-### Traditional Main Session Flow
-```
-User: "Add tests for UserService, AuthService, and OrderService"
- │
- └─→ Claude in main session:
- │
- ├─→ Loads full context (2000+ tokens)
- ├─→ Discusses approach with user
- ├─→ Implements UserService tests
- ├─→ Shows code, waits for approval
- ├─→ Implements AuthService tests
- ├─→ Shows code, waits for approval
- ├─→ Implements OrderService tests
- └─→ Total: ~8000 tokens, $0.032
-
-Main Session Cost:
- Context load: 2000 tokens
- Discussion: 1500 tokens
- Implementation: 4500 tokens
- ────────────────────────────
- Total: 8000 tokens → $0.032
-```
-
-### Delegation Flow (Token Optimized)
-```
-User: "Add tests for UserService, AuthService, and OrderService"
- │
- └─→ ccs-delegator agent:
- │
- ├─→ Analyzes: 3 similar tasks → Batch delegate
- │
- ├─→ Execute 3 delegations:
- │ │
- │ ├─→ ccs glm -p "add tests for UserService"
- │ │ └─→ Cost: $0.0015 (500 tokens)
- │ │
- │ ├─→ ccs glm -p "add tests for AuthService"
- │ │ └─→ Cost: $0.0015 (500 tokens)
- │ │
- │ └─→ ccs glm -p "add tests for OrderService"
- │ └─→ Cost: $0.0015 (500 tokens)
- │
- └─→ Total: ~1500 tokens, $0.0045
-
-Delegation Cost:
- Task 1 (GLM): 500 tokens → $0.0015
- Task 2 (GLM): 500 tokens → $0.0015
- Task 3 (GLM): 500 tokens → $0.0015
- ────────────────────────────────────────
- Total: 1500 tokens → $0.0045
-
-Savings: $0.032 - $0.0045 = $0.0275 (86% reduction) ⚡
-```
-
----
-
-## Integration Points Summary
-
-```
-┌─────────────────────────────────────────────────────────────┐
-│ Integration Points │
-├─────────────────────────────────────────────────────────────┤
-│ │
-│ 1. bin/ccs.js (lines 501-507) │
-│ └─→ Detects -p flag → Routes to DelegationHandler │
-│ │
-│ 2. bin/delegation/delegation-handler.js (NEW) │
-│ └─→ Orchestrates delegation flow │
-│ │
-│ 3. bin/delegation/headless-executor.js (EXISTING) │
-│ └─→ Spawns claude -p with enhanced flags │
-│ │
-│ 4. bin/delegation/session-manager.js (EXISTING) │
-│ └─→ Persists session metadata │
-│ │
-│ 5. bin/delegation/result-formatter.js (EXISTING) │
-│ └─→ Formats ASCII box output │
-│ │
-│ 6. .claude/commands/ccs/glm.md │
-│ └─→ Executes: ccs glm -p "$ARGUMENTS" │
-│ │
-│ 7. .claude/agents/ccs-delegator.md │
-│ └─→ Proactive delegation via Task tool │
-│ │
-│ 8. .claude/skills/ccs-delegation/ │
-│ └─→ AI decision framework + technical docs │
-│ │
-└─────────────────────────────────────────────────────────────┘
-```
-
----
-
-## Effectiveness Metrics
-
-### Feature Coverage
-
-```
-✅ Stream-JSON Output Parsing
- └─→ Real-time jsonl format, extracts: session_id, cost, turns, errors
-
-✅ Real-Time Tool Visibility
- └─→ Shows: [Tool] Bash: npm install, [Tool] Write: index.html
-
-✅ Session Management
- └─→ Persists to: ~/.ccs/delegation-sessions.json
-
-✅ Multi-Turn Support
- └─→ Resume via: ccs glm:continue -p "task"
-
-✅ Cost Tracking
- └─→ Aggregates across turns, displays in USD
-
-✅ Time-Based Limits
- └─→ Default: 10min timeout with graceful SIGTERM termination
-
-✅ Permission Mode
- └─→ Default: acceptEdits (auto-approve file ops)
-
-✅ Signal Handling
- └─→ Kills child process on Ctrl+C/Esc (no orphans)
-
-✅ Slash Command Preservation
- └─→ Detects /cook, /plan in prompts, keeps at start
-
-✅ Formatted Output
- └─→ ASCII box with metadata
-```
-
-### Performance Impact
-
-```
-Metric Before After Improvement
-────────────────────────────────────────────────────────────
-Session overhead 2000 tok 500 tok 75% ↓
-Cost per simple task $0.008 $0.0015 81% ↓
-Time to result ~30s ~10s 67% ↓
-Context pollution High Zero 100% ↓
-Batch 3 tasks $0.024 $0.0045 81% ↓
-```
-
-### User Experience Flow
-
-```
-BEFORE (Manual):
- User: "Add tests for UserService"
- Claude: "I'll add tests for UserService..."
- [Generates code in main session, uses context]
- Claude: "Here are the tests..."
- User: "Now add tests for AuthService"
- Claude: "I'll add tests for AuthService..."
- [Repeats, accumulates context]
-
-AFTER (Delegated):
- User: "Add tests for UserService, AuthService, OrderService"
- Claude: "I'll delegate these similar tasks to GLM for token optimization"
- [Batch delegates via ccs-delegator agent]
- ccs glm -p "add tests for UserService" → $0.0015
- ccs glm -p "add tests for AuthService" → $0.0015
- ccs glm -p "add tests for OrderService" → $0.0015
- Claude: "All tests added. Total cost: $0.0045"
- [Main session context stays clean]
-```
-
----
-
-## Architecture Benefits
-
-### 1. Separation of Concerns
-```
-bin/ccs.js → Routing only (6 lines added)
-delegation-handler.js → Orchestration logic
-headless-executor.js → Execution engine
-session-manager.js → State persistence
-result-formatter.js → Output formatting
-```
-
-### 2. Progressive Disclosure
-```
-SKILL.md → Entry point (56 lines)
- └─→ headless-workflow.md → Technical details (155 lines)
- └─→ delegation-guidelines.md → AI decision rules (100 lines)
-```
-
-### 3. Zero Breaking Changes
-```
-ccs glm → Works as before (normal profile)
-ccs glm -p "task" → NEW: Enhanced delegation
-ccs glm:continue -p → NEW: Multi-turn support
-```
-
-### 4. Token Efficiency
-```
-Main session: Full context loaded for every task
-Delegation: Isolated execution, no context pollution
-Savings: 81% cost reduction on simple tasks
-```
-
----
-
-## Future Enhancements
-
-### Potential Improvements
-```
-1. Cost Alerts
- └─→ Warn if delegation > $1.00
-
-2. Session Analytics
- └─→ Track delegation patterns, identify high-cost tasks
-
-3. Batch Optimization
- └─→ Auto-detect batchable tasks: "add tests for all *.service.js"
-
-4. Profile Auto-Selection
- └─→ Agent chooses GLM vs Kimi based on file count
-
-5. GLMT Integration
- └─→ Complex reasoning tasks via glmt proxy + delegation
-```
-
----
-
-## Troubleshooting Flows
-
-### Common Issues
-
-```
-Issue: "No previous session found for glm"
- │
- └─→ Cause: Using :continue without initial session
- │
- └─→ Solution: Run initial task first
- ccs glm -p "initial task"
- ccs glm:continue -p "follow up"
-
-Issue: "Profile not configured for delegation"
- │
- └─→ Cause: Missing ~/.ccs/glm.settings.json
- │
- └─→ Solution: Run ccs doctor
- ccs doctor
- → Shows configuration issues
-
-Issue: "Missing prompt after -p flag"
- │
- └─→ Cause: No argument after -p
- │
- └─→ Solution: Provide prompt in quotes
- ccs glm -p "task description"
-```
-
----
-
-**Last Updated**: 2025-11-16
-**Related**: `SKILL.md`, `headless-workflow.md`, `delegation-guidelines.md`
diff --git a/docs/code-standards.md b/docs/code-standards.md
deleted file mode 100644
index 509000e7..00000000
--- a/docs/code-standards.md
+++ /dev/null
@@ -1,1515 +0,0 @@
-# CCS Code Standards
-
-## Overview
-
-This document defines the coding standards and principles for the CCS (Claude Code Switch) project. The project uses TypeScript throughout, React 19 for the UI dashboard, Vite for fast builds, and Bun as the primary package manager. Following these standards ensures consistency, maintainability, and quality across the entire codebase.
-
-## Core Principles
-
-### Design Philosophy
-
-**YAGNI** (You Aren't Gonna Need It)
-- Only implement features that are immediately needed
-- Avoid code "just in case" scenarios
-- Keep the codebase minimal and focused
-
-**KISS** (Keep It Simple, Stupid)
-- Prefer simple solutions over complex ones
-- Avoid unnecessary abstractions
-- Use straightforward, readable code
-
-**DRY** (Don't Repeat Yourself)
-- Eliminate duplicate code through consolidation
-- Create reusable functions for common operations
-- Maintain single sources of truth
-
-### Simplification Standards
-
-The recent codebase simplification (35% reduction from 1,315 to 855 lines) established these standards:
-
-1. **Consolidate duplicate logic**: Unified spawn logic in `execClaude()` function
-2. **Remove security theater**: Eliminate unnecessary validation functions
-3. **Simplify error handling**: Direct console.error instead of complex formatting
-4. **Deduplicate platform checks**: Centralize platform-specific logic
-
-## TypeScript Standards
-
-### Code Style
-
-#### File Structure
-```typescript
-// Dependencies
-import { spawn } from 'child_process';
-import path from 'path';
-import { error } from './helpers';
-
-// Constants
-const CCS_VERSION = require('../package.json').version;
-
-// Functions (grouped by responsibility)
-function mainFunction(): void {
- // Implementation
-}
-
-// Main execution
-function main(): void {
- // Implementation
-}
-
-// Run main
-main();
-```
-
-#### Function Declarations
-- Use `function` declarations for named functions
-- Use arrow functions only for anonymous functions or callbacks
-- Group related functions together
-- Place main execution logic at the bottom
-
-```javascript
-// Good
-function handleVersionCommand() {
- console.log(`CCS version ${CCS_VERSION}`);
- process.exit(0);
-}
-
-// Acceptable for callbacks
-fs.readFile(file, (err, data) => {
- if (err) return error(err.message);
- // Process data
-});
-```
-
-#### Variable Declarations
-- Use `const` for variables that won't be reassigned
-- Use `let` only when reassignment is necessary
-- Declare variables as close to usage as possible
-- Avoid `var` entirely
-
-```javascript
-// Good
-const configPath = getConfigPath();
-const claudeCli = detectClaudeCli();
-
-// Avoid
-let configPath;
-configPath = getConfigPath();
-```
-
-### Error Handling Standards
-
-#### Simple Error Reporting
-```javascript
-// Preferred - Simple and direct
-function error(message) {
- console.error(`ERROR: ${message}`);
- process.exit(1);
-}
-
-// Usage
-if (!fs.existsSync(configPath)) {
- error(`Config file not found: ${configPath}`);
-}
-```
-
-#### Avoid Complex Error Formatting
-```javascript
-// Avoid - Complex box-drawing formatting
-function showErrorBox(message) {
- console.log('╔══════════════════════════════════════╗');
- console.log('║ ERROR ║');
- console.log(`║ ${message.padEnd(34)} ║`);
- console.log('╚══════════════════════════════════════╝');
-}
-```
-
-#### Early Validation Pattern
-```javascript
-function getSettingsPath(profile) {
- // Validate early and exit fast
- const config = readConfig();
- const settingsPath = config.profiles[profile];
-
- if (!settingsPath) {
- error(`Profile '${profile}' not found. Available: ${Object.keys(config.profiles).join(', ')}`);
- }
-
- return settingsPath;
-}
-```
-
-### Process Management Standards
-
-#### Unified Spawn Logic
-```javascript
-// Consolidated spawn function - single source of truth
-function execClaude(claudeCli, args) {
- const child = spawn(claudeCli, args, {
- stdio: 'inherit',
- windowsHide: true
- });
-
- child.on('exit', (code, signal) => {
- if (signal) process.kill(process.pid, signal);
- else process.exit(code || 0);
- });
-
- child.on('error', () => {
- showClaudeNotFoundError();
- process.exit(1);
- });
-}
-```
-
-#### Security Best Practices
-- Always use arrays with `spawn()` to prevent shell injection
-- Never construct shell command strings with user input
-- Validate inputs before using them in file operations
-
-```javascript
-// Good - Safe with array arguments
-spawn(claudeCli, ['--settings', settingsPath, ...args]);
-
-// Avoid - Unsafe string concatenation
-spawn('sh', ['-c', `claude --settings ${settingsPath} ${args.join(' ')}`]);
-```
-
-### Module Organization Standards (Phase 02 Complete - 2025-11-27)
-
-#### Subsystem Directory Structure
-```
-src/
-├── ccs.ts # Main entry point (593 lines, 44.6% reduction)
-├── commands/ # Modular command handlers (Phase 02 NEW)
-│ ├── version-command.ts # Version display functionality
-│ ├── help-command.ts # Help system
-│ ├── install-command.ts # Installation workflows
-│ ├── doctor-command.ts # System diagnostics
-│ ├── sync-command.ts # Configuration synchronization
-│ └── shell-completion-command.ts # Shell completion management
-├── auth/ # Auth system modules
-├── delegation/ # Delegation system modules
-├── glmt/ # GLMT system modules
-├── management/ # Management system modules
-├── utils/ # Utility modules (expanded in Phase 02)
-│ ├── shell-executor.ts # Cross-platform execution (NEW)
-│ └── package-manager-detector.ts # Package manager detection (NEW)
-└── types/ # TypeScript type definitions
-```
-
-#### Module Dependencies
-```javascript
-// Group imports by type
-// Node.js built-ins
-const fs = require('fs');
-const path = require('path');
-const { spawn } = require('child_process');
-
-// Local modules (relative paths from current subsystem)
-const { error } = require('../utils/helpers');
-const { detectClaudeCli } = require('../utils/claude-detector');
-
-// Subsystem-specific modules
-const DelegationHandler = require('./delegation-handler');
-const SessionManager = require('./session-manager');
-```
-
-#### Exports Pattern
-```javascript
-// Clear, named exports
-module.exports = {
- getConfigPath,
- readConfig,
- getSettingsPath
-};
-
-// Class exports (delegation, management modules)
-class DelegationHandler {
- constructor() {
- // Implementation
- }
-}
-module.exports = DelegationHandler;
-
-// Avoid exports with mixed responsibilities
-module.exports = {
- getConfigPath,
- someUtilityFunction,
- anotherUnrelatedFunction
-};
-```
-
-#### Subsystem Naming Conventions (v4.x)
-- **handler**: Routing and orchestration (e.g., `delegation-handler.js`)
-- **executor**: Execution logic (e.g., `headless-executor.js`)
-- **manager**: State management (e.g., `session-manager.js`, `instance-manager.js`)
-- **validator**: Validation logic (e.g., `delegation-validator.js`)
-- **formatter**: Output formatting (e.g., `result-formatter.js`)
-- **parser**: Parsing logic (e.g., `settings-parser.js`, `sse-parser.js`)
-
-## Modular Command Architecture Standards (Phase 02, 2025-11-27)
-
-### Overview
-
-Phase 02 introduced modular command architecture that separates command handling logic from the main orchestrator. This enhances maintainability, testability, and development workflow efficiency.
-
-### Command Handler Pattern
-
-**Structure Requirements**:
-- Each command must be implemented in its own dedicated file in `src/commands/`
-- Commands must follow consistent interface pattern for type safety
-- Single responsibility principle - each module handles one command only
-
-**Interface Pattern**:
-```typescript
-interface CommandHandler {
- handle(args: string[]): Promise
-
- );
-}
-```
-
-### UI Library Standards (shadcn/ui)
-
-#### Component Usage
-- Use shadcn/ui components as base building blocks
-- Customize with Tailwind CSS classes
-- Maintain consistent design tokens
-- Support dark mode out of the box
-
-```typescript
-import { Button } from '@/components/ui/button';
-import { Card, CardContent, CardHeader, CardTitle } from '@/components/ui/card';
-
-export function SettingsPage(): JSX.Element {
- return (
- {title}
- {isLoading ? ( -Loading...
- ) : (
-
- )}
-
-
-
-
- Active
-
- )}
-
-
- {profile.name}
- {profile.isActive && ( -
- {[...Array(6)].map((_, i) => (
-
- );
-}
-
-if (error) {
- return
-
- ))}
- Failed to load profiles: {error.message}
;
-}
-
-if (!profiles || profiles.length === 0) {
- return (
-
- No profiles configured. Create your first profile to get started.
-
- );
-}
-```
-
-## UI System Patterns (Phase 5, v4.5.0+)
-
-### Central UI Module (src/utils/ui.ts)
-
-#### Initialization Pattern
-```typescript
-import { ui } from '../utils/ui';
-
-// Initialize UI (call once at startup)
-async function main() {
- await ui.init();
- // UI functions now available
-}
-```
-
-#### Semantic Color Usage
-```typescript
-// Apply semantic colors
-ui.color('Success!', 'success') // Green bold
-ui.color('Error!', 'error') // Red bold
-ui.color('Warning!', 'warning') // Yellow
-ui.color('Info', 'info') // Cyan
-ui.color('Command: ccs', 'command') // Yellow italic
-ui.color('/path/to/file', 'path') // Cyan underline
-```
-
-#### ASCII Status Indicators (NO EMOJIS)
-```typescript
-// Preferred: ASCII indicators only
-ui.ok('Installation complete') // [OK] Installation complete
-ui.fail('Build failed') // [X] Build failed
-ui.warn('Deprecated feature') // [!] Deprecated feature
-ui.info('Checking system') // [i] Checking system
-
-// NEVER use emoji (violates CLAUDE.md)
-// DON'T: console.log('✓ Success') or '❌ Error'
-```
-
-#### Box & Table Rendering
-```typescript
-// Styled boxes
-ui.box('Content here', {
- title: 'Header',
- borderColor: '#00ECFA',
- padding: 1
-});
-
-ui.errorBox('Error message', 'ERROR');
-ui.infoBox('Info content', 'INFO');
-
-// Styled tables
-ui.table([
- ['Column 1', 'Column 2'],
- ['Data 1', 'Data 2']
-], {
- head: ['Header 1', 'Header 2'],
- style: 'unicode'
-});
-```
-
-#### Spinner & Progress
-```typescript
-// Simple spinner (falls back to plain text in non-TTY)
-const spin = await ui.spinner('Processing...');
-await doWork();
-spin.succeed('Done!');
-
-// Spinner with message update
-const spin = await ui.spinner('Loading');
-spin.update('Still loading...');
-spin.succeed('Completed');
-```
-
-#### Task Lists (Listr2 Integration - Phase 5)
-```typescript
-// Create task list with intelligent renderer selection
-const ctx = await ui.taskList([
- {
- title: 'Install dependencies',
- task: async () => {
- await installDeps();
- }
- },
- {
- title: 'Build project',
- task: async () => {
- await build();
- },
- skip: () => !needsBuild() ? 'Already built' : undefined
- }
-], {
- concurrent: false // Set true for parallel execution
-});
-
-// Renderer selection (automatic):
-// - TTY: Default renderer with subtask hierarchy
-// - Non-TTY: Simple renderer (plain text)
-// - CI environment: Simple renderer
-// - Claude Code context: Fallback spinner-based
-```
-
-#### Context Detection
-```typescript
-// Check if running in interactive TTY
-if (ui.isInteractive()) {
- // Can use full spinner/box features
-}
-
-// Check if running inside Claude Code tool
-if (ui.isClaudeCodeContext()) {
- // May need fallback rendering
-}
-```
-
-### UI Compliance Rules (STRICT)
-
-**Mandatory**:
-- ALWAYS call `await ui.init()` before using any UI functions
-- ASCII status indicators ONLY: [OK], [X], [!], [i] - NO EMOJIS
-- Respect NO_COLOR and FORCE_COLOR environment variables
-- Use semantic colors (success, error, warning, info, dim, primary, secondary, command, path)
-
-**Patterns**:
-- Status messages: `ui.ok()`, `ui.fail()`, `ui.warn()`, `ui.info()`
-- Styled output: `ui.box()`, `ui.table()`, `ui.spinner()`
-- Progress: Use `ui.taskList()` for multi-step operations (Phase 5)
-- Fallback-first: Always test in non-TTY (pipe) mode
-
-**Testing**:
-```bash
-# Test in non-TTY (pipes/CI)
-ccs doctor | cat # Forces plain text output
-NO_COLOR=1 ccs doctor # Disables colors
-```
-
-## Delegation System Patterns (v4.0+, Phase 5 Enhanced)
-
-### Stream-JSON Parsing
-
-#### Tool Extraction Pattern
-```javascript
-// Real-time extraction of tool calls from stream-JSON output
-function parseToolLine(line) {
- const toolRegex = /\[Tool\]\s+(\w+)\((.*?)\)/;
- const match = line.match(toolRegex);
-
- if (match) {
- const [_, toolName, params] = match;
- return { toolName, params };
- }
-
- return null;
-}
-
-// Usage in stream parser
-childProcess.stdout.on('data', (chunk) => {
- const lines = chunk.toString().split('\n');
- for (const line of lines) {
- const tool = parseToolLine(line);
- if (tool) {
- console.log(` ${formatToolName(tool.toolName)} ${formatParams(tool.params)}`);
- }
- }
-});
-```
-
-### Session Management Pattern
-
-#### Session Persistence
-```javascript
-// Save session for continuation
-class SessionManager {
- saveSession(profile, sessionId, prompt) {
- const sessionData = {
- profile,
- sessionId,
- timestamp: new Date().toISOString(),
- prompt
- };
-
- const sessionPath = path.join(DELEGATION_SESSIONS_DIR, `${profile}-last.json`);
- fs.writeFileSync(sessionPath, JSON.stringify(sessionData, null, 2));
- }
-
- loadSession(profile) {
- const sessionPath = path.join(DELEGATION_SESSIONS_DIR, `${profile}-last.json`);
-
- if (!fs.existsSync(sessionPath)) {
- return null;
- }
-
- return JSON.parse(fs.readFileSync(sessionPath, 'utf8'));
- }
-}
-```
-
-### Headless Execution Pattern
-
-#### Spawning with Stream-JSON
-```javascript
-// Execute Claude CLI in headless mode with stream-JSON output
-function executeHeadless(claudeCli, profile, prompt, sessionId = null) {
- const args = [
- '--output-format', 'stream-json',
- '--verbose',
- '-p', prompt
- ];
-
- if (sessionId) {
- args.push('--session-id', sessionId);
- }
-
- const child = spawn(claudeCli, args, {
- stdio: ['inherit', 'pipe', 'pipe'],
- windowsHide: true
- });
-
- // Parse stdout for tools
- child.stdout.on('data', (chunk) => {
- parseStreamOutput(chunk.toString());
- });
-
- // Handle Ctrl+C
- process.on('SIGINT', () => {
- child.kill('SIGTERM');
- process.exit(130);
- });
-
- return child;
-}
-```
-
-### Result Formatting Pattern (Phase 5 Enhanced - Async)
-
-#### Async Result Formatter with UI Integration
-```typescript
-// result-formatter.ts - Phase 5 now fully async
-import { ui } from '../utils/ui';
-
-class ResultFormatter {
- // Initialize UI and format result
- static async format(result: ExecutionResult): Promise
-
-# CCS - Claude Code Switch
-
-
-
-### Claude Code用ユニバーサルAIプロファイルマネージャー
-
-**複数のClaudeアカウント、任意のAnthropic互換API、OAuthプロバイダー（Gemini、Codex、Antigravity）を瞬時に切り替え。**
-レート制限を回避し、無制限のプロファイルで継続的に作業。
-
-
- -[](LICENSE) -[]() -[](https://www.npmjs.com/package/@kaitranntt/ccs) -[](https://claudekit.cc?ref=HMNKXOHN) - -**Languages**: [English](../../README.md) · [Tiếng Việt](../vi/README.md) · [日本語](README.md) - -
-
-- -[](LICENSE) -[]() -[](https://www.npmjs.com/package/@kaitranntt/ccs) -[](https://claudekit.cc?ref=HMNKXOHN) - -**Languages**: [English](../../README.md) · [Tiếng Việt](../vi/README.md) · [日本語](README.md) - -
- -## クイックスタート - -### インストール - -**npmパッケージ（推奨）** - -**macOS / Linux / Windows** -```bash -npm install -g @kaitranntt/ccs -``` - -**主要なパッケージマネージャーすべてに対応：** - -```bash -# yarn -yarn global add @kaitranntt/ccs - -# pnpm（ディスク使用量70%削減） -pnpm add -g @kaitranntt/ccs - -# bun（30倍高速） -bun add -g @kaitranntt/ccs -``` - -
-
- -> [!WARNING] -> **これらのインストーラーは非推奨であり、将来のバージョンで削除されます。** -> 現在は npm インストールに自動リダイレクトされます。npm を直接使用してください。 - -**macOS / Linux** -```bash -curl -fsSL ccs.kaitran.ca/install | bash -``` - -**Windows PowerShell** -```powershell -irm ccs.kaitran.ca/install | iex -``` - -**注**: スクリプトは非推奨の警告を表示し、Node.jsが利用可能な場合は自動的にnpmインストールを実行します。 - -
-
-[!] 非推奨: ネイティブシェルインストーラー（レガシー）
- -- -> [!WARNING] -> **これらのインストーラーは非推奨であり、将来のバージョンで削除されます。** -> 現在は npm インストールに自動リダイレクトされます。npm を直接使用してください。 - -**macOS / Linux** -```bash -curl -fsSL ccs.kaitran.ca/install | bash -``` - -**Windows PowerShell** -```powershell -irm ccs.kaitran.ca/install | iex -``` - -**注**: スクリプトは非推奨の警告を表示し、Node.jsが利用可能な場合は自動的にnpmインストールを実行します。 - -
- -### 設定（自動作成） - -**CCSはインストール時に自動的に設定を作成します**（npm postinstallスクリプト経由）。 - -**~/.ccs/config.json**: -```json -{ - "profiles": { - "glm": "~/.ccs/glm.settings.json", - "glmt": "~/.ccs/glmt.settings.json", - "kimi": "~/.ccs/kimi.settings.json", - "default": "~/.claude/settings.json" - } -} -``` - -
-
-
-
- -Claude CLIが標準以外の場所（Dドライブ、カスタムディレクトリ）にインストールされている場合は、`CCS_CLAUDE_PATH`を設定してください： - -```bash -# Unix/Linux/macOS -export CCS_CLAUDE_PATH="/path/to/claude" - -# Windows PowerShell -$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe" -``` - -**参照**: [トラブルシューティングガイド](./docs/en/troubleshooting.md#claude-cli-in-non-standard-location) 詳細な設定手順 - -
-
-カスタムClaude CLIパス
-
-- -Claude CLIが標準以外の場所（Dドライブ、カスタムディレクトリ）にインストールされている場合は、`CCS_CLAUDE_PATH`を設定してください： - -```bash -# Unix/Linux/macOS -export CCS_CLAUDE_PATH="/path/to/claude" - -# Windows PowerShell -$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe" -``` - -**参照**: [トラブルシューティングガイド](./docs/en/troubleshooting.md#claude-cli-in-non-standard-location) 詳細な設定手順 - -
-
-
-
- -**Windowsユーザー**: 本物のシンボリックリンクで高速な動作と即時同期を得るために開発者モードを有効にしてください： - -1. **設定** → **プライバシーとセキュリティ** → **開発者向け** を開く -2. **開発者モード** を有効にする -3. CCSを再インストール: `npm install -g @kaitranntt/ccs` - -**警告**: 開発者モードなしの場合、CCSは自動的にディレクトリコピーにフォールバック（動作しますが、プロファイル間の即時同期はありません） - -
-
-Windowsシンボリックリンクサポート（開発者モード）
-
-- -**Windowsユーザー**: 本物のシンボリックリンクで高速な動作と即時同期を得るために開発者モードを有効にしてください： - -1. **設定** → **プライバシーとセキュリティ** → **開発者向け** を開く -2. **開発者モード** を有効にする -3. CCSを再インストール: `npm install -g @kaitranntt/ccs` - -**警告**: 開発者モードなしの場合、CCSは自動的にディレクトリコピーにフォールバック（動作しますが、プロファイル間の即時同期はありません） - -
- -### 最初の切り替え - -> [!IMPORTANT] -> **代替モデルを使用する前に、設定ファイルでAPIキーを更新してください：** -> -> - **GLM**: `~/.ccs/glm.settings.json`を編集してZ.AI Coding Plan APIキーを追加 -> - **GLMT**: `~/.ccs/glmt.settings.json`を編集してZ.AI Coding Plan APIキーを追加 -> - **Kimi**: `~/.ccs/kimi.settings.json`を編集してKimi APIキーを追加 - -
- -**並列ワークフロー：計画 + 実行** - -```bash -# Terminal 1 - 計画（Claude Sonnet） -ccs "認証とレート制限付きREST APIの計画" - -# Terminal 2 - 実行（GLM、コスト最適化） -ccs glm "計画からユーザー認証エンドポイントを実装" -``` - -
-
- -```bash -# Kimi - 安定した思考サポート -ccs kimi "トレードオフ分析付きキャッシュ戦略の設計" - -# GLMT - 実験的（詳細は下記参照） -ccs glmt "推論ステップ付き複雑なアルゴリズムのデバッグ" -``` - -**注**: GLMTは実験的で不安定です。詳細については下記の[GLM with Thinking (GLMT)](#glm-with-thinking-glmt)セクションを参照してください。 - -
-
-思考モデル（Kimi & GLMT）
- -- -```bash -# Kimi - 安定した思考サポート -ccs kimi "トレードオフ分析付きキャッシュ戦略の設計" - -# GLMT - 実験的（詳細は下記参照） -ccs glmt "推論ステップ付き複雑なアルゴリズムのデバッグ" -``` - -**注**: GLMTは実験的で不安定です。詳細については下記の[GLM with Thinking (GLMT)](#glm-with-thinking-glmt)セクションを参照してください。 - -
- -## 開発者の日常的な課題 - -
-
-### **切り替えを停止。調整を開始。**
-
-**セッション制限がフロー状態を殺すべきではありません。**
-
-
-実装に深く集中しています。コンテキストが読み込まれました。解決策が結晶化しています。-その後: 🔴 _"使用制限に達しました。"_ - -**モチベーションが失われました。コンテキストが失われました。生産性が崩壊しました。** - -## **解決策：並列ワークフロー** - -
-
-
-❌ 古い方法： 制限に達した時に切り替える（反応的）
- -### 現在のワークフロー： -- **14時:** 機能開発、ゾーン状態 -- **15時:** 🔴 使用制限に達した -- **15:05:** 作業停止、`~/.claude/settings.json`を編集 -- **15:15:** アカウント切り替え、コンテキストが失われる -- **15:30:** フロー状態に戻ろうと試みる -- **16時:** ついに生産性が回復 - -- **結果:** 1時間失われ、モチベーションが破壊され、不満が蓄積 - -
-
-
-## あなたの道を選択
-
-✨ 新しい方法： 最初から並列で実行（主導的） - 推奨
- -### 新しいワークフロー： -- **14時:** **ターミナル1:** `ccs "APIアーキテクチャを計画"` → 戦略的思考（Claude Pro） -- **14時:** **ターミナル2:** `ccs glm "エンドポイントを実装"` → コード実行（GLM） -- **15時:** まだ開発継続、割れなし -- **16時:** フロー状態達成、生産性急上昇 -- **17時:** 機能が完了、コンテキスト維持 - -- **結果:** ダウムタイムなし、継続的生産性、不満減少 - -### 💰 **価値提案:** -- **設定:** 既存のClaude Pro + GLM Lite（費用対効果の高い追加） -- **価値:** 1時間/日 × 20労働日 = 20時間/月を回収 -- **ROI:** 開発時間は設定コスト以上の価値がある -- **現実:** オーバーヘッドより速く出荷 - -
-
-
-予算重視： GLMのみ
- -- **最適:** 費用意識の高い開発、基本的なコード生成 -- **使用法:** 費用効果の高いAI支援のために`ccs glm`を直接使用 -- **現実:** Claudeアクセスなし、多くのコーディングタスクに対応可能 -- **設定:** GLM APIキーのみ、非常に手頃 - -
-
-
-✨ 日々の開発に推奨： 1 Claude Pro + 1 GLM Lite
- -- **最適:** 日々のコードデリバリー、真剣な開発作業 -- **使用法:** `ccs`で計画 + `ccs glm`で実行（並列ワークフロー） -- **現実:** ほとんどの開発者にとって能力と費用の完璧なバランス -- **価値:** セッション制限に達せず、継続的生産性 - -
-
-
-パワーユーザー： 複数のClaude Pro + GLM Pro
- -- **最適:** 重い作業量、並行プロジェクト、ソロ開発 -- **解放:** セッション・週次制限を決して枯渇させない -- **ワークフロー:** 3+以上のターミナルで専門タスクを同時実行 - -
-
-
----
-
-## 手動切り替えではなくCCSを使う理由は？
-
-プライバシー重視： 仕事/個人の分離
- -- **必要時:** 仕事と個人AIコンテキストの厳格な分離 -- **設定:** `ccs auth create work` + `ccs auth create personal` -- **注意:** 高度な機能 - ほとんどのユーザーには不要 - -
-
-**CCSは「午後3時に制限に達したら切り替える」ことではありません。**
-
-## **それは最初から並列で実行することです。**
-
-
-
-### コアな違い
-
-| **手動切り替え** | **CCSオーケストレーション** |
-|:---|:---|
-| 🔴 制限達成 → 作業停止 → 設定ファイル編集 → 再起動 | ✅ 最初から異なるモデルで複数ターミナルを実行 |
-| 😰 コンテキストロスとフロー状態中断 | 😌 コンテキスト維持での継続的生産性 |
-| 📝 逐次的タスク処理 | ⚡ 並列ワークフロー（計画 + 実行を同時に） |
-| 🛠️ ブロックされた時の反応的問題解決 | 🎯 ブロックを防ぐ主導的ワークフロー設計 |
-
-### CCSが提供するもの
-
-- **ゼロコンテキスト切り替え:** 割れずにフロー状態を維持
-- **並列生産性:** 1ターミナルで戦略計画、もう1つでコード実行
-- **即座アカウント管理:** 1コマンド切り替え、設定ファイル編集不要
-- **仕事と生活の分離:** ログアウトせずにコンテキストを分離
-- **クロスプラットフォーム一貫性:** macOS、Linux、Windowsで同じスムーズな体験
-
-- -## アーキテクチャ - -### プロファイルタイプ - -**設定ベース**: GLM, GLMT, Kimi, default -- 設定ファイルを指す`--settings`フラグを使用 -- GLMT: 思考モードサポートの埋め込みプロキシ - -**アカウントベース**: work, personal, team -- 分離されたインスタンスに`CLAUDE_CONFIG_DIR`を使用 -- `ccs auth create
- -## 使用例 - -### 基本的な切り替え - -```bash -ccs # Claudeサブスクリプション（デフォルト） -ccs glm # GLM（コスト最適化） -ccs kimi # Kimi（思考サポート付き） -``` - -### マルチアカウント設定 - -```bash -# アカウントを作成 -ccs auth create work -ccs auth create personal -``` - -**別々のターミナルで同時に実行：** - -```bash -# Terminal 1 - 業務用 -ccs work "機能を実装" - -# Terminal 2 - 個人用（同時） -ccs personal "コードレビュー" -``` - -### ヘルプとバージョン - -```bash -ccs --version # バージョンを表示 -ccs --help # すべてのコマンドとオプションを表示 -``` - -
- -## GLM with Thinking (GLMT) - -> [!CAUTION] -> ### 本番環境未対応 - 実験的機能 -> -> **GLMTは実験的で広範なデバッグが必要です**： -> - ストリーミングとツールサポートはまだ開発中 -> - 予期せぬエラー、タイムアウト、不完全な応答が発生する可能性 -> - 頻繁なデバッグと手動介入が必要 -> - **重要なワークフローや本番使用には推奨されません** -> -> **GLM Thinkingの代替案**: **CCR hustle**と**BedollaのTransformer**（[ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/)）を通じて、より安定した実装を検討してください。 - -> [!IMPORTANT] -> GLMTはnpmインストールが必要です（`npm install -g @kaitranntt/ccs`）。ネイティブシェルバージョンでは利用できません（Node.js HTTPサーバーが必要）。 - -
- -> [!NOTE] -> ### 謝辞：GLMTを可能にした基盤 -> -> **CCSのGLMT実装は、[@Bedolla](https://github.com/Bedolla)の画期的な仕事に存在を負っています**。彼は[Claude Code Router (CCR)](https://github.com/musistudio/claude-code-router)とZ.AIの推論能力をブリッジする[最初の統合](https://github.com/Bedolla/ZaiTransformer/)を作成しました。 -> -> ZaiTransformer以前、誰もZ.AIの思考モードとClaude Codeのワークフローを正常に統合できませんでした。Bedollaの仕事は単なる有用なものではなく、**基盤的**でした。彼のリクエスト/レスポンストランスフォーメーションアーキテクチャ、思考モード制御メカニズム、埋め込みプロキシ設計の実装は、GLMTの設計に直接インスピレーションを与え、可能にしました。 -> -> **ZaiTransformerの先駆的な仕事なしでは、GLMTは現在の形では存在しませんでした。** GLMTの思考能力から利益を得る場合は、Claude Codeエコシステムでの先駆的な仕事をサポートするために[ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/)にスターを付けてください。 - -
- -
-
-
-
- -
-
-GLM vs GLMT 比較
-
-- -
-
-| 機能 | GLM (`ccs glm`) | GLMT (`ccs glmt`) |
-|:--------|:----------------|:------------------|
-| **エンドポイント** | Anthropic互換 | OpenAI互換 |
-| **思考** | なし | 実験的（reasoning_content） |
-| **ツールサポート** | 基本的 | **不安定（v3.5+）** |
-| **MCPツール** | 制限あり | **バグあり（v3.5+）** |
-| **ストリーミング** | 安定 | **実験的（v3.4+）** |
-| **TTFB** | <500ms | <500ms（時々）、2-10秒+（頻繁） |
-| **使用例** | 信頼性の高い作業 | **デバッグ実験のみ** |
-
-
-
-- -
-
-
-
- -**GLMTはMCPツールと関数呼び出しを試行：** - -- **双方向トランスフォーメーション**: Anthropicツール ↔ OpenAI形式（不安定） -- **MCP統合**: MCPツールが時々実行（多くの場合XMLガベージを出力） -- **ストリーミングツール呼び出し**: リアルタイムツール呼び出し（クラッシュしない場合） -- **後方互換**: 既存の思考サポートを破壊する可能性 -- **設定が必要**: 頻繁な手動デバッグが必要 - -
-
-ツールサポート（v3.5） - 実験的
-
-- -**GLMTはMCPツールと関数呼び出しを試行：** - -- **双方向トランスフォーメーション**: Anthropicツール ↔ OpenAI形式（不安定） -- **MCP統合**: MCPツールが時々実行（多くの場合XMLガベージを出力） -- **ストリーミングツール呼び出し**: リアルタイムツール呼び出し（クラッシュしない場合） -- **後方互換**: 既存の思考サポートを破壊する可能性 -- **設定が必要**: 頻繁な手動デバッグが必要 - -
-
-
-
- -**GLMTは増分推論コンテンツ配信でリアルタイムストリーミングを試行：** - -- **デフォルト**: ストリーミング有効（動作時TTFB <500ms） -- **自動フォールバック**: エラーにより頻繁にバッファモードに切り替え -- **思考パラメータ**: Claude CLI `thinking`パラメータが時々動作 - - `thinking.type`と`budget_tokens`を無視する場合 - - 優先順位: CLIパラメータ > メッセージタグ > デフォルト（破壊されていない場合） - -**ステータス**: Z.AI（テスト済み、ツール呼び出しが頻繁に破壊、継続的なデバッグが必要） - -
-
-ストリーミングサポート（v3.4） - しばしば失敗
-
-- -**GLMTは増分推論コンテンツ配信でリアルタイムストリーミングを試行：** - -- **デフォルト**: ストリーミング有効（動作時TTFB <500ms） -- **自動フォールバック**: エラーにより頻繁にバッファモードに切り替え -- **思考パラメータ**: Claude CLI `thinking`パラメータが時々動作 - - `thinking.type`と`budget_tokens`を無視する場合 - - 優先順位: CLIパラメータ > メッセージタグ > デフォルト（破壊されていない場合） - -**ステータス**: Z.AI（テスト済み、ツール呼び出しが頻繁に破壊、継続的なデバッグが必要） - -
-
-
-
- -1. CCSがlocalhostに埋め込みHTTPプロキシを生成（クラッシュしない場合） -2. プロキシがAnthropic形式 → OpenAI形式への変換を試行（多くの場合失敗） -3. Anthropicツール → OpenAI関数呼び出し形式への変換を試行（バグあり） -4. 推論パラメータとツールを付けてZ.AIに転送（タイムアウトしない場合） -5. `reasoning_content` → 思考ブロックへの変換を試行（部分的または破壊） -6. OpenAI `tool_calls` → Anthropic `tool_use` ブロックへの変換を試行（XMLガベージが一般的） -7. 思考とツール呼び出しが時々Claude Code UIに表示（破壊されていない場合） - -
-
-動作原理（動作時）
-
-- -1. CCSがlocalhostに埋め込みHTTPプロキシを生成（クラッシュしない場合） -2. プロキシがAnthropic形式 → OpenAI形式への変換を試行（多くの場合失敗） -3. Anthropicツール → OpenAI関数呼び出し形式への変換を試行（バグあり） -4. 推論パラメータとツールを付けてZ.AIに転送（タイムアウトしない場合） -5. `reasoning_content` → 思考ブロックへの変換を試行（部分的または破壊） -6. OpenAI `tool_calls` → Anthropic `tool_use` ブロックへの変換を試行（XMLガベージが一般的） -7. 思考とツール呼び出しが時々Claude Code UIに表示（破壊されていない場合） - -
-
-
-
- -**制御タグ**: -- `` - 推論ブロックの有効/無効（デフォルト: On）
-- `` - 推論深度の制御（非推奨 - Z.AIはバイナリ思考のみサポート）
-
-**思考キーワード**（不安定なアクティベーション）:
-- `think` - 時々推論を有効化（低労力）
-- `think hard` - 時々推論を有効化（中労力）
-- `think harder` - 時々推論を有効化（高労力）
-- `ultrathink` - 最大推論深度を試行（多くの場合破壊）
-
-
-
-制御タグとキーワード
-
-- -**制御タグ**: -- `
-
-
-
- -**GLMT機能**（すべて実験的）: -- 強制的な英語出力強制（時々動作） -- ランダムな思考モードアクティベーション（予測不可能） -- 頻繁なバッファモードへのフォールバック付きストリーミング試行 - -**一般**: -- `CCS_DEBUG_LOG=1` - デバッグファイルロギングを有効化 -- `CCS_CLAUDE_PATH=/path/to/claude` - カスタムClaude CLIパス - -
-
-環境変数
-
-- -**GLMT機能**（すべて実験的）: -- 強制的な英語出力強制（時々動作） -- ランダムな思考モードアクティベーション（予測不可能） -- 頻繁なバッファモードへのフォールバック付きストリーミング試行 - -**一般**: -- `CCS_DEBUG_LOG=1` - デバッグファイルロギングを有効化 -- `CCS_CLAUDE_PATH=/path/to/claude` - カスタムClaude CLIパス - -
-
-
-
- -```bash -# GLMT設定を編集 -nano ~/.ccs/glmt.settings.json -``` - -Z.AI APIキーを設定（コーディングプランが必要）： - -```json -{ - "env": { - "ANTHROPIC_AUTH_TOKEN": "your-z-ai-api-key" - } -} -``` - -
-
-APIキー設定
-
-- -```bash -# GLMT設定を編集 -nano ~/.ccs/glmt.settings.json -``` - -Z.AI APIキーを設定（コーディングプランが必要）： - -```json -{ - "env": { - "ANTHROPIC_AUTH_TOKEN": "your-z-ai-api-key" - } -} -``` - -
-
-
-
- -**v3.4 保護制限**: - -| 制限 | 値 | 目的 | -|:------|:------|:--------| -| **SSEバッファ** | イベントあたり最大1MB | バッファオーバーフローを防止 | -| **コンテンツバッファ** | ブロックあたり最大10MB | 思考/テキストブロックを制限 | -| **コンテンツブロック** | メッセージあたり最大100 | DoS攻撃を防止 | -| **リクエストタイムアウト** | 120秒 | ストリーミングとバッファの両方 | - -
-
-セキュリティ制限（DoS保護）
-
-- -**v3.4 保護制限**: - -| 制限 | 値 | 目的 | -|:------|:------|:--------| -| **SSEバッファ** | イベントあたり最大1MB | バッファオーバーフローを防止 | -| **コンテンツバッファ** | ブロックあたり最大10MB | 思考/テキストブロックを制限 | -| **コンテンツブロック** | メッセージあたり最大100 | DoS攻撃を防止 | -| **リクエストタイムアウト** | 120秒 | ストリーミングとバッファの両方 | - -
-
-
-
- -**詳細ロギングを有効化**: -```bash -ccs glmt --verbose "your prompt" -``` - -**デバッグファイルロギングを有効化**: -```bash -export CCS_DEBUG_LOG=1 -ccs glmt --verbose "your prompt" -# ログ: ~/.ccs/logs/ -``` - -**GLMTデバッグ**: -```bash -# 詳細ロギングでストリーミングステータスと推論詳細を表示 -ccs glmt --verbose "test" -``` - -**推論コンテンツを確認**: -```bash -cat ~/.ccs/logs/*response-openai.json | jq '.choices[0].message.reasoning_content' -``` - -**トラブルシューティング**: -- **存在しない場合**: Z.AI APIの問題（キー、アカウントステータスを確認） -- **存在する場合**: トランスフォーメーションの問題（`response-anthropic.json`を確認） - -
-
-デバッグ
-
-- -**詳細ロギングを有効化**: -```bash -ccs glmt --verbose "your prompt" -``` - -**デバッグファイルロギングを有効化**: -```bash -export CCS_DEBUG_LOG=1 -ccs glmt --verbose "your prompt" -# ログ: ~/.ccs/logs/ -``` - -**GLMTデバッグ**: -```bash -# 詳細ロギングでストリーミングステータスと推論詳細を表示 -ccs glmt --verbose "test" -``` - -**推論コンテンツを確認**: -```bash -cat ~/.ccs/logs/*response-openai.json | jq '.choices[0].message.reasoning_content' -``` - -**トラブルシューティング**: -- **存在しない場合**: Z.AI APIの問題（キー、アカウントステータスを確認） -- **存在する場合**: トランスフォーメーションの問題（`response-anthropic.json`を確認） - -
- -## アンインストール - -
-
-
-
- -```bash -# npm -npm uninstall -g @kaitranntt/ccs - -# yarn -yarn global remove @kaitranntt/ccs - -# pnpm -pnpm remove -g @kaitranntt/ccs - -# bun -bun remove -g @kaitranntt/ccs -``` - -
-
-パッケージマネージャー
-
-- -```bash -# npm -npm uninstall -g @kaitranntt/ccs - -# yarn -yarn global remove @kaitranntt/ccs - -# pnpm -pnpm remove -g @kaitranntt/ccs - -# bun -bun remove -g @kaitranntt/ccs -``` - -
-
-
-
- -```bash -# macOS / Linux -curl -fsSL ccs.kaitran.ca/uninstall | bash - -# Windows PowerShell -irm ccs.kaitran.ca/uninstall | iex -``` - -
-
-公式アンインストーラー
-
-- -```bash -# macOS / Linux -curl -fsSL ccs.kaitran.ca/uninstall | bash - -# Windows PowerShell -irm ccs.kaitran.ca/uninstall | iex -``` - -
- -## 🎯 哲学 - -- **YAGNI**: 「念のため」の機能は追加しない -- **KISS**: シンプルなbash、複雑さなし -- **DRY**: 単一の情報源（設定） - -## 📖 ドキュメント - -**[docs/](./docs/)の完全なドキュメント**: -- [インストールガイド](./docs/en/installation.md) -- [設定](./docs/en/configuration.md) -- [使用例](./docs/en/usage.md) -- [システムアーキテクチャ](./docs/system-architecture.md) -- [GLMT制御メカニズム](./docs/glmt-controls.md) -- [トラブルシューティング](./docs/en/troubleshooting.md) -- [貢献](./CONTRIBUTING.md) - -## 🤝 貢献 - -貢献を歓迎します！詳細については[貢献ガイド](./CONTRIBUTING.md)をご覧ください。 - -## Star History - -
-
-
-
-
-
-
-## ライセンス
-
-CCSは[MITライセンス](LICENSE)の下でライセンスされています。
-
-
-
-**レート制限に頻繁に遭遇する開発者のために ❤️ を込めて作成**
-
-[⭐ このリポジトリにスター](https://github.com/kaitranntt/ccs) | [🐛 問題を報告](https://github.com/kaitranntt/ccs/issues) | [📖 ドキュメントを読む](./docs/en/)
-
-
\ No newline at end of file
diff --git a/docs/project-overview-pdr.md b/docs/project-overview-pdr.md
deleted file mode 100644
index d52826a4..00000000
--- a/docs/project-overview-pdr.md
+++ /dev/null
@@ -1,497 +0,0 @@
-# CCS Project Overview and Product Development Requirements (PDR)
-
-## Executive Summary
-
-CCS (Claude Code Switch) is a TypeScript-based CLI tool that enables instant profile switching between Claude, GLM, Kimi, Gemini, Codex, and Antigravity (and any OAuth-based models) models. It features a version pin mechanism for precise model control and a unified configuration system for seamless management across all providers. The project features a modern React 19 web dashboard with real-time WebSocket integration, comprehensive TypeScript architecture, and cross-platform support. Current architecture includes:
-
-- **TypeScript Core**: 43 source files with 100% type coverage
-- **React 19 Dashboard**: Modern UI with Vite, shadcn/ui, and real-time features
-- **AI Delegation System**: Headless execution with stream-JSON output
-- **Cross-Platform**: Native support for macOS, Linux, and Windows
-- **163 total files**: ~8,000 lines of TypeScript code
-
-## Product Vision
-
-### Mission Statement
-Provide developers with instant, zero-downtime switching between diverse AI models, ensuring precise version control, optimizing for cost, performance, and rate limit management while maintaining a seamless workflow experience.
-
-### Core Value Proposition
-- **Instant Switching**: One command to change AI models and providers without file editing
-- **Zero Downtime**: Never interrupt development workflow during model or provider switches
-- **Version Pinning**: Lock to specific model versions for consistent behavior
-- **Cost Optimization**: Use the right model/provider for each task automatically
-- **Developer Experience**: Maintain familiar Claude CLI interface with enhanced capabilities and multi-provider support
-
-## Product Development Requirements (PDR)
-
-### Functional Requirements
-
-#### FR-001: Profile Management
-**Requirement**: System shall support instant switching between multiple AI model profiles
-- **Priority**: High
-- **Acceptance Criteria**:
- - Switch profiles with single command (`ccs glm`, `ccs`)
- - Maintain profile state until explicitly changed
- - Support unlimited profile configurations
- - Automatic profile detection from command arguments
-
-#### FR-002: Configuration Management
-**Requirement**: System shall provide automatic configuration management
-- **Priority**: High
-- **Acceptance Criteria**:
- - Auto-create configuration during installation
- - Support custom configuration paths via environment variables
- - Validate configuration file format and existence
- - Provide clear error messages for configuration issues
-
-#### FR-003: Claude CLI Integration
-**Requirement**: System shall seamlessly integrate with official Claude CLI
-- **Priority**: High
-- **Acceptance Criteria**:
- - Pass all arguments transparently to Claude CLI
- - Support all Claude CLI features and flags
- - Maintain identical user experience to native Claude CLI
- - Auto-detect Claude CLI installation location
-
-#### FR-004: Cross-Platform Compatibility
-**Requirement**: System shall work identically across all supported platforms
-- **Priority**: High
-- **Acceptance Criteria**:
- - Support macOS (Intel and Apple Silicon)
- - Support Linux distributions
- - Support Windows (PowerShell and Git Bash)
- - Consistent behavior and error handling across platforms
-
-#### FR-005: Special Command Support
-**Requirement**: System shall support special meta-commands for management
-- **Priority**: Medium
-- **Acceptance Criteria**:
- - `ccs --version` displays version and installation location
- - `ccs --help` shows usage information
- - **WIP**: `ccs --install` integrates with Claude Code commands (testing incomplete)
- - **WIP**: `ccs --uninstall` removes Claude Code integration (testing incomplete)
-
-#### FR-006: Error Handling
-**Requirement**: System shall provide clear, actionable error messages
-- **Priority**: Medium
-- **Acceptance Criteria**:
- - Validate configuration file existence and format
- - Detect Claude CLI availability and report issues
- - Provide suggestions for resolving common problems
- - Maintain consistent error message format
-
-#### FR-007: AI-Powered Delegation System
-**Requirement**: System shall enable headless AI task delegation with real-time tool tracking
-- **Priority**: High
-- **Acceptance Criteria**:
- - Execute Claude CLI in headless mode with `-p` flag
- - Parse stream-JSON output for real-time tool visibility
- - Track 13+ Claude Code tools (Bash, Read, Write, Edit, Glob, Grep, etc.)
- - Support session continuation (`:continue` suffix)
- - Display cost and duration statistics
- - Handle Ctrl+C signal properly (kill child processes)
-
-#### FR-008: .claude/ Directory Symlinking
-**Requirement**: System shall selectively symlink .claude/ directories for data sharing
-- **Priority**: Medium
-- **Acceptance Criteria**:
- - Symlink shared data: commands/, skills/, agents/
- - Keep profile-specific data isolated: settings.json, sessions/, todolists/, logs/
- - Windows fallback to directory copying when symlinks unavailable
- - Non-invasive installation (never modify ~/.claude/settings.json)
- - Idempotent installation (safe to run multiple times)
-
-#### FR-009: Shell Completion
-**Requirement**: System shall provide comprehensive shell completion across 4 shells
-- **Priority**: Low
-- **Acceptance Criteria**:
- - Support Bash, Zsh, Fish, PowerShell
- - Color-coded categories (profiles, commands, flags)
- - Profile-aware completions (glm, glmt, kimi, work, personal)
- - Easy installation via `--shell-completion` flag
- - Show installation instructions per shell
-
-#### FR-010: Diagnostics and Maintenance
-**Requirement**: System shall provide comprehensive health diagnostics
-- **Priority**: Medium
-- **Acceptance Criteria**:
- - `ccs doctor`: Validate installation, profiles, symlinks, API keys
- - `ccs sync`: Fix broken symlinks and directory structure
- - `ccs update`: Check for newer versions with smart notifications
- - Color-coded status indicators ([OK], [!], [X])
- - Actionable recommendations for issues
-
-#### FR-011: Web Dashboard UI
-**Requirement**: System shall provide a modern web dashboard for profile and configuration management
-- **Priority**: High
-- **Acceptance Criteria**:
- - Profile management interface with active/inactive states
- - Interactive command builder with search and categorization
- - Performance metrics visualization with trend indicators
- - Responsive design supporting mobile, tablet, and desktop
- - Real-time updates via WebSocket integration
- - Dark mode support out of the box
-
-### Non-Functional Requirements
-
-#### NFR-001: Performance
-**Requirement**: System shall execute with minimal overhead
-- **Priority**: High
-- **Acceptance Criteria**:
- - Profile switching completes in < 100ms
- - Startup time < 50ms for any command
- - Memory footprint < 10MB during execution
- - No perceptible delay compared to native Claude CLI
-
-#### NFR-002: Reliability
-**Requirement**: System shall maintain 99.9% uptime during normal operations
-- **Priority**: High
-- **Acceptance Criteria**:
- - Handle edge cases gracefully without crashes
- - Maintain functionality across system reboots
- - Recover gracefully from temporary system issues
- - No memory leaks or resource exhaustion
-
-#### NFR-003: Security
-**Requirement**: System shall follow security best practices
-- **Priority**: High
-- **Acceptance Criteria**:
- - No shell injection vulnerabilities in process execution
- - Validate file paths to prevent traversal attacks
- - Use secure process spawning with argument arrays
- - No storage of sensitive credentials or API keys
-
-#### NFR-004: Maintainability
-**Requirement**: System shall be easy to maintain and extend
-- **Priority**: Medium
-- **Acceptance Criteria**:
- - Code complexity maintained at manageable levels
- - Comprehensive test coverage (>90%)
- - Clear documentation and code comments
- - Modular architecture supporting future enhancements
-
-#### NFR-005: Usability
-**Requirement**: System shall provide excellent developer experience
-- **Priority**: Medium
-- **Acceptance Criteria**:
- - Intuitive command structure matching CLI conventions
- - Clear help documentation and usage examples
- - Minimal learning curve for existing Claude CLI users
- - Consistent behavior across all use cases
-
-## Technical Architecture
-
-### System Components
-
-#### Core Subsystems (v4.3.2)
-1. **Main Entry Point** (`bin/ccs.js` ~800 lines): Command parsing, profile routing, delegation detection
-2. **Auth System** (`bin/auth/` ~800 lines): Multi-account management (create, list, delete, switch)
-3. **Delegation System** (`bin/delegation/` ~1,200 lines): AI-powered task delegation with stream-JSON
-4. **GLMT System** (`bin/glmt/` ~700 lines): Thinking mode proxy and transformation
-5. **Management System** (`bin/management/` ~600 lines): Config, instance, profile, shared data management
-6. **Utilities** (`bin/utils/` ~1,500 lines): Symlink manager, validators, update checker, completion
-7. **.claude/ Integration** (`~/.ccs/shared/`): Symlinked commands, skills, agents directories
-
-#### v4.0-4.3.2 Major Enhancements
-- **AI Delegation** (v4.0): Headless execution with stream-JSON output, session continuation
-- **Selective Symlinking** (v4.1): Share .claude/ directories (commands, skills, agents)
-- **Shell Completion** (v4.1.4): 4 shells supported with color-coded categories
-- **Diagnostics** (v4.2): Doctor, sync, update commands for health checks
-- **Stream-JSON Parser** (v4.3): Real-time tool tracking during delegation
-
-### Data Flow (v4.3.2)
-
-**Settings-based profiles (glm, kimi, glmt)**:
-```mermaid
-graph LR
- USER[ccs glm "task"] --> PARSE[Parse Args]
- PARSE --> DETECT[ProfileDetector]
- DETECT --> CONFIG[Read config.json]
- CONFIG --> EXEC[execClaude with --settings]
- EXEC --> CLAUDE[Claude CLI]
-```
-
-**Delegation execution (v4.0+)**:
-```mermaid
-graph LR
- USER[ccs glm -p "task"] --> DELEGATE[DelegationHandler]
- DELEGATE --> HEADLESS[HeadlessExecutor]
- HEADLESS --> STREAM[Parse stream-JSON]
- STREAM --> FORMAT[ResultFormatter]
- FORMAT --> SESSION[SessionManager save]
-```
-
-**Account-based profiles (work, personal)**:
-```mermaid
-graph LR
- USER[ccs work "task"] --> PARSE[Parse Args]
- PARSE --> DETECT[ProfileDetector]
- DETECT --> INSTANCE[InstanceManager.ensureInstance]
- INSTANCE --> EXEC[execClaude with CLAUDE_CONFIG_DIR]
- EXEC --> CLAUDE[Claude CLI reads from instance]
-```
-
-**Evolution Flow Comparison**:
-- **v2.x**: Login → Encrypt → Store → Decrypt → Copy → Execute (6 steps)
-- **v3.0**: Create instance → Login → Execute (3 steps, 50% reduction)
-- **v4.x**: Add delegation routing → Stream-JSON parsing → Session persistence (extended capabilities)
-
-### Configuration Architecture (v4.3.2)
-
-**Settings-based Config** (`~/.ccs/config.json`):
-```json
-{
- "profiles": {
- "glm": "~/.ccs/glm.settings.json",
- "glmt": "~/.ccs/glmt.settings.json",
- "kimi": "~/.ccs/kimi.settings.json",
- "default": "~/.claude/settings.json"
- }
-}
-```
-
-**Shared Data Architecture** (v4.1+):
-```
-~/.ccs/shared/ # Symlinked to instance .claude/ directories
-├── commands/ # Slash commands (shared across profiles)
-├── skills/ # Agent skills (shared across profiles)
-└── agents/ # Agent configs (shared across profiles)
-```
-
-**Account Profile Registry** (`~/.ccs/profiles.json`):
-```json
-{
- "version": "2.0.0",
- "profiles": {
- "work": {
- "type": "account",
- "created": "2025-11-09T10:00:00.000Z",
- "last_used": "2025-11-09T15:30:00.000Z"
- }
- },
- "default": "work"
-}
-```
-
-**v3.0 Schema Simplification**:
-- **Removed fields**: `vault`, `subscription`, `email` (not needed for login-per-profile)
-- **Kept fields**: `type`, `created`, `last_used` (essential metadata only)
-- **Rationale**: Credentials live in instance directories, no vault needed
-
-**Instance Directory Structure** (v4.1+ with symlinking):
-```
-~/.ccs/instances/work/
-├── .claude/
-│ ├── commands@ → ~/.ccs/shared/commands/ # Symlink (v4.1+)
-│ ├── skills@ → ~/.ccs/shared/skills/ # Symlink (v4.1+)
-│ ├── agents@ → ~/.ccs/shared/agents/ # Symlink (v4.1+)
-│ ├── settings.json # Profile-specific (isolated)
-│ ├── sessions/ # Profile-specific (isolated)
-│ ├── todolists/ # Profile-specific (isolated)
-│ └── logs/ # Profile-specific (isolated)
-├── .anthropic/ # SDK config
-└── .credentials.json # Login credentials (managed by Claude CLI)
-```
-
-- **Environment Override**: `CCS_CLAUDE_PATH` - Custom Claude CLI path
-- **Auto-Creation**: Configuration generated automatically during installation
-
-## Implementation Standards
-
-### Code Quality Standards
-- **YAGNI Principle**: Only implement features immediately needed
-- **KISS Principle**: Maintain simplicity over complexity
-- **DRY Principle**: Eliminate code duplication
-- **Test Coverage**: >90% coverage for all critical paths
-- **Documentation**: Clear code comments and external documentation
-
-### Development Workflow
-1. **Feature Development**: Implement following coding standards
-2. **Testing**: Comprehensive unit and integration tests
-3. **Documentation**: Update relevant documentation
-4. **Quality Review**: Code review against standards checklist
-5. **Release**: Version management and distribution
-
-### Platform Support Matrix
-| Platform | Version Support | Testing Coverage |
-|----------|----------------|------------------|
-| macOS | 10.15+ | Full |
-| Linux | Ubuntu 18.04+, CentOS 7+ | Full |
-| Windows | 10+ (PowerShell, Git Bash) | Full |
-
-## Quality Assurance
-
-### Testing Strategy
-- **Unit Tests**: Individual module functionality
-- **Integration Tests**: Cross-module interaction
-- **Platform Tests**: OS-specific behavior validation
-- **Edge Case Tests**: Error conditions and boundary cases
-- **Performance Tests**: Resource usage and response time
-
-### Quality Metrics
-- **Code Coverage**: >90% line coverage
-- **Complexity**: Maintain cyclomatic complexity < 10 per function
-- **Performance**: Startup time < 50ms, memory < 10MB
-- **Reliability**: <0.1% error rate in normal operations
-
-## Deployment and Distribution
-
-### Distribution Channels
-- **npm Package**: Primary distribution channel (`@kaitranntt/ccs`)
-- **Direct Install**: Platform-specific install scripts
-- **GitHub Releases**: Source code and binary distributions
-
-### Installation Methods
-1. **npm Package** (Recommended): `npm install -g @kaitranntt/ccs`
-2. **Direct Install**: `curl -fsSL ccs.kaitran.ca/install | bash`
-3. **Windows PowerShell**: `irm ccs.kaitran.ca/install | iex`
-
-### Auto-Configuration Process
-1. **Package Installation**: npm or direct script execution
-2. **Post-install Hook**: Automatic configuration creation
-3. **Path Setup**: Add to system PATH when needed
-4. **Validation**: Verify Claude CLI availability
-5. **Ready State**: System ready for profile switching
-
-## Success Metrics
-
-### v4.3.2 Achievement Metrics
-- **Delegation System**: AI-powered task execution with stream-JSON output
-- **Tool Tracking**: 13+ Claude Code tools supported
-- **Session Persistence**: `:continue` support for follow-up tasks
-- **Shell Completion**: 4 shells supported (Bash, Zsh, Fish, PowerShell)
-- **Diagnostics**: Doctor, sync, update commands for health checks
-- **Symlinking**: Selective .claude/ directory sharing (commands, skills, agents)
-
-### Phase 1 UI Achievement Metrics (2024-12-12)
-- **Core Components**: 5 foundational UI components implemented
- - ProfileCard: Individual profile display with state management
- - ProfileDeck: Grid container with loading/error states
- - CommandBuilder: Interactive CLI tool with search/filter
- - ValueMetrics: Performance visualization with trends
- - HubFooter: Navigation footer with version info
-- **Responsive Design**: Mobile-first approach with Tailwind CSS
-- **Component Architecture**: Modular, reusable component patterns
-- **Mock Data Integration**: Development-ready with clear data flow patterns
-
-### Adoption Metrics
-- **Download Count**: npm package downloads per month
-- **Installation Success Rate**: >95% successful installations
-- **User Retention**: Monthly active users
-- **Platform Distribution**: Usage across supported platforms
-- **Delegation Usage**: Percentage of users utilizing `-p` flag
-- **API Key Configuration**: Rate of users configuring GLM/Kimi/GLMT keys
-
-### Performance Metrics (v4.3.2)
-- **Profile Creation**: ~5-10ms (instance directory creation only)
-- **Profile Activation**: ~5-10ms (no decryption overhead)
-- **Delegation Startup**: <500ms (stream-JSON initialization)
-- **Response Time**: Minimal overhead, direct Claude CLI execution
-- **Error Rate**: <0.1% in normal operations
-- **Reliability**: 99.9% uptime during normal operations
-
-### Quality Metrics
-- **Test Coverage**: >90% for all critical paths
-- **Bug Reports**: Number and severity of reported issues
-- **Fix Time**: Average time to resolve reported issues
-- **User Satisfaction**: Feedback and ratings
-
-## Risk Management
-
-### Technical Risks
-- **Claude CLI Changes**: API changes in official CLI
- - **Mitigation**: Maintain abstraction layer, monitor changes
-- **Platform Compatibility**: OS-specific issues
- - **Mitigation**: Comprehensive testing, CI/CD across platforms
-- **Dependency Issues**: npm package or system dependency problems
- - **Mitigation**: Minimal dependencies, regular testing
-
-### Business Risks
-- **Competition**: Similar tools emerging
- - **Mitigation**: Focus on simplicity and reliability
-- **User Adoption**: Slow adoption rates
- - **Mitigation**: Clear documentation, easy installation
-- **Maintenance Burden**: Ongoing maintenance costs
- - **Mitigation**: Simplified codebase, automated testing
-
-## Future Roadmap
-
-See [docs/project-roadmap.md](./project-roadmap.md) for detailed version history and future plans.
-
-### Completed (v4.3.2)
-- ✅ **AI Delegation System** (v4.0): Headless execution with stream-JSON
-- ✅ **Selective Symlinking** (v4.1): Share .claude/ directories
-- ✅ **Shell Completion** (v4.1.4): 4 shells with color-coded categories
-- ✅ **Diagnostics** (v4.2): Doctor, sync, update commands
-- ✅ **Session Continuation** (v4.3): `:continue` support
-- ✅ **Vault Removal** (v3.0): Login-per-profile model
-- ✅ **Platform Parity** (v3.0): Unified macOS/Linux/Windows behavior
-
-### Active Development (v4.4-v4.5)
-- **Delegation Improvements**: MCP tool integration, SQLite session storage
-- **Performance Optimization**: Model selection based on task complexity
-- **Enhanced Diagnostics**: Automated troubleshooting recommendations
-
-### UI Development (Phase 1 Complete - 2024-12-12)
-- ✅ **Core Components**: ProfileCard, ProfileDeck, CommandBuilder, ValueMetrics, HubFooter
-- ✅ **Responsive Design**: Mobile-first implementation with Tailwind CSS
-- ✅ **Component Architecture**: Established patterns for future development
-- **Phase 2 (Planned)**: API integration, data persistence, advanced components
-- **Phase 3 (Future)**: Charts/graphs, real-time updates, enhanced interactivity
-
-### Future Considerations (v5.0+)
-- **AI-Powered Features**: Automatic task classification, intelligent model selection
-- **Enterprise Features**: Team profile sharing, usage analytics dashboard
-- **Ecosystem Expansion**: Plugin system for custom models, CI/CD integration
-
-## Compliance and Legal
-
-### Licensing
-- **MIT License**: Permissive open-source license
-- **Third-party Dependencies**: All dependencies use compatible licenses
-- **Attribution**: Proper attribution for all used components
-
-### Privacy
-- **Data Collection**: No personal data collection or transmission
-- **Local Processing**: All processing happens locally
-- **Configuration Privacy**: User configurations remain private
-
-### Security
-- **Code Review**: Regular security reviews and audits
-- **Dependency Management**: Regular updates and vulnerability scanning
-- **Secure Distribution**: Signed packages and secure distribution channels
-
-## Conclusion
-
-The CCS project demonstrates successful iterative evolution balancing simplification with enhanced capabilities:
-
-### Evolution Summary
-- **v2.x**: Vault-based credential encryption (~1,700 LOC)
-- **v3.0**: Vault removal, login-per-profile (~1,100 LOC, 40% reduction)
-- **v4.0-4.3.2**: AI delegation, .claude/ symlinking, stream-JSON (~8,477 LOC with enhanced features)
-
-### v4.3.2 Architectural Benefits
-1. **AI-Powered Delegation**: Headless execution with real-time tool tracking
-2. **Selective Symlinking**: Shared .claude/ data (commands, skills, agents) across profiles
-3. **Enhanced Diagnostics**: Doctor, sync, update commands for health checks
-4. **Stream-JSON Parsing**: Real-time visibility into Claude Code tool usage
-5. **Session Persistence**: Continue delegation sessions with `:continue` suffix
-6. **Shell Completion**: 4 shells supported with color-coded categories
-7. **Platform Parity**: Unified behavior across macOS/Linux/Windows
-
-### Key Strengths (v4.3.2)
-- **Modular Architecture**: Clear subsystem separation (auth, delegation, glmt, management, utils)
-- **Non-Invasive Installation**: Never modifies ~/.claude/settings.json
-- **Idempotent Operations**: Safe to run installation/sync multiple times
-- **Cross-Platform Compatibility**: Unified behavior, Windows symlink fallback
-- **Developer Experience**: Familiar Claude CLI interface with AI delegation
-- **Maintainability**: Modular design, clear responsibilities
-- **Performance**: Minimal overhead, direct CLI execution
-- **Unified Configuration**: Centralized management for all AI models and providers
-
-### Breaking Changes (v3.x → v4.x)
-- **Zero Breaking Changes**: v4.x fully backward compatible with v3.x
-- **New Features**: Delegation, symlinking, diagnostics added without breaking existing workflows
-- **Migration**: No migration required from v3.x to v4.x
-
-The project is well-positioned for future growth with a solid architectural foundation, comprehensive AI delegation capabilities, and enhanced developer experience. The v4.x architecture provides a sustainable basis for continued enhancement while maintaining core principles of simplicity, reliability, and performance.
\ No newline at end of file
diff --git a/docs/project-roadmap.md b/docs/project-roadmap.md
deleted file mode 100644
index 052a51ea..00000000
--- a/docs/project-roadmap.md
+++ /dev/null
@@ -1,862 +0,0 @@
-# CCS Project Roadmap
-
-## Project Overview
-
-CCS (Claude Code Switch) is a TypeScript-based CLI tool with a modern React 19 dashboard for instant switching between multiple AI models (Claude Sonnet 4.5, GLM 4.6, GLMT, Kimi). The project features real-time WebSocket integration, comprehensive profile management, and cross-platform support. It enables developers to maintain continuous productivity by running parallel workflows with different AI models, avoiding rate limits and context switching.
-
-### Core Value Proposition
-- **Zero Downtime**: Instant profile switching without breaking flow state
-- **Modern UI**: React 19 dashboard with real-time updates and WebSocket integration
-- **Cost Optimization**: 81% cost savings through intelligent delegation to GLM/Kimi
-- **Parallel Workflows**: Strategic planning with Claude + cost-effective execution with GLM
-- **Cross-Platform**: Unified experience on macOS, Linux, and Windows
-- **Type Safety**: 100% TypeScript coverage with zero `any` types
-
-## Current Status
-
-### Version Information
-- **Current Version**: 4.5.0 (React Dashboard Integration Complete)
-- **Release Status**: Production-ready with modern React UI
-- **Build Status**: ✅ Working (npm run build → dist/ccs.js)
-- **UI Build Status**: ✅ Working (cd ui && bun run build → dist/)
-- **Test Status**: ✅ All tests passing (39/39 tests)
-- **Cross-Platform**: ✅ Windows/macOS/Linux
-- **Code Quality**: ✅ ESLint strictness upgrade completed
-- **Code Architecture**: ✅ TypeScript-based with modular components
-- **React Dashboard**: ✅ Full React 19 integration with Vite and shadcn/ui
-- **Real-time Features**: ✅ WebSocket integration for live updates
-
-### React Dashboard Integration (v4.5.0)
-
-**✅ Complete Modern UI Implementation**
-
-CCS now features a comprehensive React 19 dashboard providing a modern web interface for profile management and system monitoring.
-
-**Technology Stack**:
-- **React 19**: Latest version with concurrent features and hooks
-- **TypeScript**: Full type safety across the entire UI codebase
-- **Vite**: Fast build tool with HMR and optimized production builds
-- **shadcn/ui**: Modern component library built on Radix UI primitives
-- **TanStack Query**: Powerful server state management with caching
-- **Tailwind CSS**: Utility-first styling for rapid development
-
-**Dashboard Features**:
-- ✅ **Real-time Updates**: WebSocket integration for live profile status
-- ✅ **Profile Management**: Visual configuration of API profiles (GLM, GLMT, Kimi)
-- ✅ **CLIProxy Integration**: OAuth provider setup and management
-- ✅ **Account Management**: Multi-account profile switching
-- ✅ **Health Monitoring**: System diagnostics dashboard
-- ✅ **Settings Panel**: Global configuration interface
-- ✅ **Dark Mode**: Theme switching support
-- ✅ **Responsive Design**: Mobile-friendly interface
-
-**Technical Implementation**:
-- **163 Total Files**: Comprehensive codebase with modular architecture
-- **~8,000 Lines**: Well-organized TypeScript code
-- **Component Library**: Reusable UI components with consistent design
-- **API Integration**: Seamless backend communication
-- **Build Optimization**: Code splitting and lazy loading
-- **Accessibility**: WCAG-compliant components
-
-### TypeScript Conversion Summary
-
-**✅ Conversion Completed: 100% (43 files)**
-
-The CCS project has been fully converted from JavaScript to TypeScript, delivering enhanced type safety, improved developer experience, and better maintainability.
-
-**Migration Statistics**:
-- **Source Files**: 43 TypeScript files converted
-- **Lines of Code**: ~8,000 lines of TypeScript code
-- **Type Coverage**: 100% with zero `any` types
-- **Build System**: Full TypeScript compilation pipeline
-- **Type Definitions**: Comprehensive type definitions in `src/types/`
-
-**Converted Components**:
-- ✅ **Core System** (`src/ccs.ts`) - Main entry point
-- ✅ **Commands** (`src/commands/`) - Modular command handlers
-- ✅ **Authentication** (`src/auth/`) - Profile management and commands
-- ✅ **CLIProxy** (`src/cliproxy/`) - OAuth provider integration
-- ✅ **Delegation** (`src/delegation/`) - AI-powered task delegation system
-- ✅ **GLMT** (`src/glmt/`) - GLM with Thinking support
-- ✅ **Management** (`src/management/`) - System diagnostics and instance management
-- ✅ **Utils** (`src/utils/`) - Cross-platform utilities and helpers
-- ✅ **Types** (`src/types/`) - Complete type definitions
-
-## Recent Achievements
-
-### TypeScript Conversion Benefits (v4.4.0)
-
-#### Type Safety & Reliability
-- **Zero `any` types**: Complete type coverage across the entire codebase
-- **Compile-time error detection**: Catches bugs before runtime
-- **Interface contracts**: Clear API boundaries and data structures
-- **Exhaustive type checking**: Eliminates entire classes of common errors
-
-#### Enhanced Developer Experience
-- **IDE support**: Full IntelliSense autocomplete and navigation
-- **Refactoring safety**: Type-safe code modifications and renames
-- **Self-documenting code**: Type definitions serve as living documentation
-- **Better debugging**: Clear type information in debuggers and stack traces
-
-#### Maintainability Improvements
-- **Code navigation**: Easy "go to definition" across the entire codebase
-- **Impact analysis**: Clear understanding of where types are used
-- **API documentation**: Types define precise interfaces for all components
-- **Future-proofing**: Easier to add new features without breaking existing code
-
-#### Architecture Enhancements
-- **Modular type system**: Comprehensive type definitions in `src/types/`
-- **Strict configuration**: TypeScript strict mode enabled for maximum safety
-- **Build pipeline**: Automated compilation with source maps and declarations
-- **Cross-platform consistency**: Types ensure consistent behavior across platforms
-
-### Phase 01: ESLint Strictness Upgrade Complete ✅
-
-**Completion Date**: 2025-11-27
-**Status**: SUCCESS - All quality gates passing
-
-#### Enhanced Code Quality Standards
-
-**ESLint Rules Upgraded** (`eslint.config.mjs`):
-- ✅ `@typescript-eslint/no-unused-vars`: `warn` → `error`
-- ✅ `@typescript-eslint/no-explicit-any`: `warn` → `error`
-- ✅ `@typescript-eslint/no-non-null-assertion`: `warn` → `error`
-
-#### Validation Results
-- **TypeScript Compilation**: ✅ Zero type errors
-- **ESLint Validation**: ✅ Zero violations (previously 3 warning rules)
-- **Test Suite**: ✅ 39/39 tests passing
-- **Manual Testing**: ✅ All critical commands working
-- **Code Coverage**: ✅ 10,487 lines of TypeScript code analyzed
-
-#### Quality Improvements Achieved
-1. **Type Safety Enhancement**: Zero tolerance for unused variables, explicit any types, and non-null assertions
-2. **Code Quality Enforcement**: Stricter linting prevents entire categories of potential bugs
-3. **Maintainability Boost**: Cleaner, more predictable codebase with enforced standards
-4. **Zero Breaking Changes**: All functionality preserved, enhanced reliability only
-
-#### Next Phase Readiness
-- **Phase 02 Status**: ✅ COMPLETED (2025-11-27)
-- **Focus**: CCS monolithic split (ccs.ts 1071 lines → 593 lines, 44.6% reduction)
-- **Goal**: ✅ Achieved - 7 modular command handlers created for improved maintainability
-- **Results**: All validation gates pass, code review: EXCELLENT
-
-**Phase 03 Status**: ✅ COMPLETED (2025-12-03)
-- **Focus**: Beta channel implementation with npm tag switching
-- **Goal**: ✅ Achieved - Full beta channel support with stability warnings
-- **Results**: All validation gates pass, comprehensive user guidance implemented
-
-### Phase 02: CCS Split Refactoring Complete ✅
-
-**Completion Date**: 2025-11-27
-**Status**: SUCCESS - All objectives achieved
-
-### Phase 03: Beta Channel Implementation Complete ✅
-
-**Completion Date**: 2025-12-03
-**Status**: SUCCESS - Beta channel fully implemented
-
-#### Beta Channel Features Delivered
-
-**NPM Tag Switching System**:
-- ✅ Implemented `fetchVersionFromNpmTag()` function for tag-specific version queries
-- ✅ Added support for `@latest` and `@dev` npm tags
-- ✅ Seamless switching between stable and beta channels
-
-**Enhanced Update Command**:
-- ✅ `--beta` flag implementation with target tag detection
-- ✅ Stability warnings for beta channel installations:
- - "[!] Installing from @dev channel (unstable)"
- - "[!] Not recommended for production use"
- - "[!] Use `ccs update` (without --beta) to return to stable"
-- ✅ Combined flag support: `--force --beta` for force reinstall from beta
-
-**Installation Method Validation**:
-- ✅ npm installations: Full beta channel support
-- ✅ Direct installer installations: Clear error messages with migration guidance
-- ✅ Automatic detection of installation method
-- ✅ Graceful fallback to stable channel for unsupported methods
-
-#### Technical Implementation Details
-
-**Core Changes**:
-- `src/utils/update-checker.ts`: Added `fetchVersionFromNpmTag()` and targetTag parameter
-- `src/commands/update-command.ts`: Enhanced with beta flag parsing and validation
-- `src/ccs.ts`: Updated to pass beta flag to update handler
-
-**User Experience Improvements**:
-- Clear warnings about beta channel stability
-- One-command return to stable channel
-- Helpful error messages with migration instructions
-- Seamless channel switching without data loss
-
-#### Validation Results
-- **TypeScript Compilation**: ✅ Zero type errors
-- **ESLint Validation**: ✅ Zero violations
-- **Manual Testing**: ✅ All beta channel scenarios working
-- **Error Handling**: ✅ Comprehensive error messages for edge cases
-- **Documentation**: ✅ Updated across all relevant files
-
-#### Modular Command Architecture Implementation
-
-**Command Handler Modules Created:**
-- ✅ `src/commands/version-command.ts` (3.0KB) - Version display functionality
-- ✅ `src/commands/help-command.ts` (4.9KB) - Comprehensive help system
-- ✅ `src/commands/install-command.ts` (957B) - Install/uninstall operations
-- ✅ `src/commands/doctor-command.ts` (415B) - System diagnostics
-- ✅ `src/commands/sync-command.ts` (1.0KB) - Configuration synchronization
-- ✅ `src/commands/shell-completion-command.ts` (2.1KB) - Shell completion management
-
-**New Utility Modules:**
-- ✅ `src/utils/shell-executor.ts` (1.5KB) - Cross-platform shell execution
-- ✅ `src/utils/package-manager-detector.ts` (3.8KB) - Package manager detection
-
-#### Refactoring Metrics Achieved
-
-**File Size Reduction:**
-- **src/ccs.ts**: 1,071 → 593 lines (**44.6% reduction**)
-- **Main file focus**: Now contains only routing logic + profile detection + GLMT proxy
-- **Maintainability**: Enhanced through modular command separation
-
-**Code Organization:**
-- **Before**: 1 monolithic file handling all commands
-- **After**: 7 focused command handlers + 2 utility modules
-- **Benefits**: Easier testing, better code navigation, focused responsibilities
-
-#### Architecture Improvements
-
-**Enhanced Modularity:**
-- **Command Handlers**: Each command isolated in dedicated module
-- **Utility Functions**: Cross-platform shell execution and package management
-- **Clean Separation**: Main file focuses on orchestration only
-- **Type Safety**: All new modules maintain 100% TypeScript compliance
-
-**Maintainability Benefits:**
-- **Single Responsibility**: Each module has focused purpose
-- **Easy Testing**: Command handlers can be tested independently
-- **Code Navigation**: Developers can quickly locate specific functionality
-- **Future Enhancements**: New commands can be added without touching main file
-
-#### Validation Results
-- **TypeScript Compilation**: ✅ Zero type errors
-- **ESLint Validation**: ✅ Zero violations
-- **Test Suite**: ✅ All tests passing
-- **Functionality**: ✅ 100% feature preservation
-- **Performance**: ✅ No degradation, minor improvement due to smaller main file
-
-#### Code Quality Assessment
-- **Review Status**: EXCELLENT
-- **Architecture**: Clean modular design
-- **Type Safety**: Comprehensive TypeScript coverage
-- **Documentation**: Well-documented interfaces and functions
-- **Future-Proofing**: Scalable architecture for additional commands
-
-### Configuration Architecture Improvements
-- **Shared Settings**: v4.4 introduces unified `settings.json` sharing across profiles
-- **Plugin Support**: Enhanced shared directory structure with plugin support
-- **Error Handling**: Improved error management with typed error codes
-- **Shell Completion**: PowerShell compatibility improvements
-
-### Phase 05: Bootstrap Passthrough Verification Complete ✅
-
-**Completion Date**: 2025-12-03 16:40
-**Status**: SUCCESS - Bootstrap scripts handle argument passthrough correctly
-
-#### Implementation Summary
-
-The bootstrap script verification phase confirmed that both Unix and Windows bootstrap scripts already correctly pass all command-line arguments to the Node.js implementation. No code changes were required.
-
-#### Technical Verification
-
-**Bash Bootstrap (`lib/ccs`)**:
-- Line 32: `exec npx "$PACKAGE" "$@"`
-- `"$@"` correctly expands to all positional parameters, preserving quotes
-- Successfully passes `--force`, `--beta`, and combined flags
-
-**PowerShell Bootstrap (`lib/ccs.ps1`)**:
-- Lines 5-8: Parameter capture with `ValueFromRemainingArguments=$true`
-- Line 38: `& npx $PACKAGE @RemainingArgs`
-- `@RemainingArgs` uses splatting to pass all captured arguments
-- Successfully passes `--force`, `--beta`, and combined flags
-
-#### Key Achievements
-
-1. **Zero Code Changes**: Existing implementation already supports new flags
-2. **Cross-Platform Consistency**: Both Unix and Windows handle arguments identically
-3. **No Breaking Changes**: Existing functionality remains unaffected
-4. **Security Preserved**: Arguments passed through without shell expansion risks
-
-#### Validation Results
-
-- **Argument Passthrough**: ✅ `--force` flag reaches Node.js implementation
-- **Beta Support**: ✅ `--beta` flag reaches Node.js implementation
-- **Combined Flags**: ✅ `--force --beta` combination works correctly
-- **Existing Commands**: ✅ All existing commands continue to work
-- **Error Handling**: ✅ Invalid flags properly rejected by Node.js layer
-
-#### Next Steps
-
-Ready to proceed with:
-- Phase 06: Comprehensive test suite implementation
-- Phase 07: Documentation updates for new features
-
-## Current Architecture
-
-### Updated Architecture Diagram
-
-```mermaid
-%%{init: {'theme': 'base', 'themeVariables': { 'primaryColor': '#1f2937', 'edgeLabelBackground':'#374151', 'clusterBkg':'#374151'}}}%%
-graph TB
- subgraph "TypeScript Source Layer (Phase 02 Modular)"
- MAIN[src/ccs.ts - Main Entry - 593 lines]
- CMDS[src/commands/ - Modular Commands]
- AUTH[src/auth/ - Authentication]
- DELEGATE[src/delegation/ - AI Delegation]
- GLMT[src/glmt/ - GLM Thinking]
- MGMT[src/management/ - System Mgmt]
- UTILS[src/utils/ - Cross-Platform Utils]
- TYPES[src/types/ - Type Definitions]
-
- subgraph "Command Handlers"
- VER[version-command.ts]
- HELP[help-command.ts]
- INSTALL[install-command.ts]
- DOCTOR[doctor-command.ts]
- SYNC[sync-command.ts]
- SHELL[shell-completion-command.ts]
- end
-
- subgraph "New Utils"
- SHELL_EXEC[shell-executor.ts]
- PKG_MGR[package-manager-detector.ts]
- end
- end
-
- %% Modular connections
- MAIN --> CMDS
- CMDS --> VER
- CMDS --> HELP
- CMDS --> INSTALL
- CMDS --> DOCTOR
- CMDS --> SYNC
- CMDS --> SHELL
- UTILS --> SHELL_EXEC
- UTILS --> PKG_MGR
-
- subgraph "Build Output Layer"
- DIST[dist/ccs.js - Compiled Output]
- DECL[dist/ccs.d.ts - Type Declarations]
- MAP[dist/ccs.js.map - Source Maps]
- end
-
- subgraph "Native Integration Layer"
- BASH[lib/ccs - Bash Implementation]
- PS[lib/ccs.ps1 - PowerShell Implementation]
- SHELLS[Shell Completion Scripts]
- end
-
- subgraph "Runtime Environment"
- CLAUDE[Claude CLI]
- CONFIG[Configuration Files]
- SHARED[~/.ccs/shared/]
- INSTANCES[Isolated Instances]
- end
-
- MAIN --> DIST
- AUTH --> DIST
- DELEGATE --> DIST
- GLMT --> DIST
- MGMT --> DIST
- UTILS --> DIST
- TYPES --> DECL
-
- DIST --> CLAUDE
- BASH --> CLAUDE
- PS --> CLAUDE
- SHELLS --> CLAUDE
- CONFIG --> CLAUDE
- SHARED --> CLAUDE
- INSTANCES --> CLAUDE
-```
-
-### TypeScript Architecture Benefits
-
-#### Type System Organization
-```
-src/types/
-├── cli.ts # CLI interface definitions
-├── config.ts # Configuration type schemas
-├── delegation.ts # Delegation system types
-├── glmt.ts # GLMT-specific types
-├── utils.ts # Utility function types
-└── index.ts # Central type exports
-```
-
-#### Component Type Safety
-- **Authentication**: Type-safe profile management and command execution
-- **Delegation**: Structured delegation workflows with typed result handling
-- **GLMT**: Type-safe transformation pipelines and proxy management
-- **Management**: Typed diagnostics and instance management
-- **Utils**: Cross-platform utilities with consistent interfaces
-
-#### Build Pipeline
-- **TypeScript Compiler**: `tsc` with strict configuration and comprehensive checks
-- **Declaration Generation**: Automatic `.d.ts` file generation for API documentation
-- **Source Maps**: Debug-friendly mapping back to TypeScript source files
-- **Incremental Builds**: Fast rebuilds with TypeScript incremental compilation
-
-## Development Roadmap
-
-### Next Development Phases
-
-#### Phase 1: TypeScript Foundation Stabilization (v4.4.1 - v4.4.5)
-**Timeline**: Immediate - 4 weeks
-
-**Priorities**:
-1. **Type System Refinement**
- - Migrate remaining `any` types (target: zero any types)
- - Add generic constraints for better type inference
- - Implement branded types for enhanced type safety
- - Create utility types for common patterns
-
-2. **Build System Enhancement**
- - Add TypeScript ESLint with strict rules
- - Implement pre-commit type checking
- - Add automated type coverage reporting
- - Enhance development build performance
-
-3. **Testing Infrastructure**
- - Migrate tests to TypeScript
- - Add type-safe test utilities
- - Implement mock type definitions
- - Create type-safe test data fixtures
-
-#### Phase 2: Advanced Delegation System (v4.5.0)
-**Timeline**: 1-2 months
-
-**TypeScript-Driven Features**:
-1. **Smart Delegation Routing**
- - Type-based task classification system
- - Intelligent profile selection algorithms
- - Cost estimation with typed calculation models
- - Performance optimization through type-aware caching
- - Recently implemented significant UI improvements for analytics dashboard, enhancing data presentation.
-
-2. **Enhanced Session Management**
- - Type-safe session persistence with serialization
- - Advanced session resumption with state restoration
- - Multi-turn conversation context management
- - Session analytics with typed metrics
-
-3. **Plugin System Foundation**
- - Type-safe plugin API definitions
- - Plugin discovery and registration system
- - Sandboxed plugin execution environment
- - Community plugin marketplace infrastructure
-
-#### Phase 3: Enterprise Features (v4.6.0)
-**Timeline**: 2-3 months
-
-**Enterprise-Grade TypeScript Features**:
-1. **Team Profile Management**
- - Type-safe role-based access control (RBAC)
- - Centralized configuration management
- - Audit logging with typed event structures
- - Compliance reporting with structured data formats
-
-2. **Advanced Analytics Dashboard**
- - Real-time usage metrics with typed data streams
- - Cost tracking and budgeting with financial types
- - Team productivity analytics with KPI types
- - Performance monitoring with alert types
-
-3. **CI/CD Integration**
- - Type-safe GitHub Actions workflows
- - Automated testing with type validation
- - Deployment pipelines with type checks
- - Infrastructure as Code with TypeScript
-
-#### Phase 4: UI Enhancements - Sidebar Redesign (v4.7.0)
-**Timeline**: 1-2 months (TBD)
-
-**Priorities**:
-1. **Sidebar Redesign Implementation**
- - Implement new sidebar component with modern UI/UX principles.
- - Ensure responsiveness across different screen sizes.
- - Integrate with existing routing and navigation logic.
- - Improve accessibility features for the sidebar.
-
-### Technical Debt Resolution
-
-#### Immediate Priorities (v4.4.1)
-1. **GLMT System Refactoring**
- - Transform GLMT proxy with typed request/response handling
- - Implement type-safe SSE parsing with error recovery
- - Add typed transformation pipelines for API compatibility
- - Create comprehensive GLMT test suites with type safety
-
-2. **Error Handling Enhancement**
- - Implement typed error hierarchies with inheritance
- - Create structured error reporting with localization
- - Add automatic error recovery with typed retry logic
- - Develop error analytics with classification systems
-
-3. **Performance Optimization**
- - Profile and optimize hot paths with typed benchmarks
- - Implement memory-efficient data structures with generics
- - Add lazy loading patterns with type-safe initialization
- - Create performance monitoring with typed metrics
-
-#### Medium-term Improvements (v4.5.0)
-1. **Cross-Platform Consistency**
- - Unify behavior across platforms with typed abstractions
- - Implement platform-specific optimizations with type guards
- - Create comprehensive cross-platform test suites
- - Add platform-specific feature detection
-
-2. **API Evolution**
- - Design backward-compatible API evolution strategies
- - Implement API versioning with type migration paths
- - Create deprecation warnings with upgrade guidance
- - Add automated API compatibility testing
-
-## Future Vision (v5.0+)
-
-### Next-Generation Architecture
-
-#### AI-Powered TypeScript Development
-1. **Intelligent Code Generation**
- - TypeScript-aware AI code completion
- - Automatic type inference from usage patterns
- - Smart refactoring suggestions with type analysis
- - Code quality recommendations based on type metrics
-
-2. **Adaptive Type System**
- - Dynamic type generation based on runtime behavior
- - Machine learning-assisted type predictions
- - Automatic interface extraction from APIs
- - Smart type narrowing with statistical analysis
-
-#### Enterprise TypeScript Platform
-1. **Advanced Team Collaboration**
- - Real-time collaborative TypeScript editing
- - Shared type libraries across projects
- - Automated type governance policies
- - Team-wide type consistency enforcement
-
-2. **Comprehensive Analytics**
- - Type usage analytics and optimization recommendations
- - Code quality metrics with type safety scores
- - Developer productivity tracking with type metrics
- - Technical debt measurement with type analysis
-
-#### Ecosystem Integration
-1. **Package Management**
- - Type-safe dependency management
- - Automated vulnerability scanning with type analysis
- - Smart version resolution with compatibility checking
- - Package quality assessment with type metrics
-
-2. **Tool Integration**
- - TypeScript-native IDE extensions
- - Enhanced debugging with type information
- - Performance profiling with type insights
- - Automated testing with type-driven test generation
-
-## Release Notes
-
-### Version 4.5.0 - Beta Channel Implementation Complete
-**Release Date**: 2025-12-03
-
-#### Major Features
-- ✅ **Beta Channel Support**: Full implementation of npm tag switching between `@latest` and `@dev`
-- ✅ **Enhanced Update Command**: `--beta` flag with stability warnings and validation
-- ✅ **Force Reinstall Support**: `--force` flag to reinstall current version and fix corrupted installs
-- ✅ **Combined Flag Support**: `--force --beta` for force reinstalling beta versions
-- ✅ **Installation Method Detection**: Differential support for npm vs direct installations
-- ✅ **User Safety Features**: Clear warnings and migration guidance for beta usage
-- ✅ **Version Fetching Enhancement**: Tag-specific queries with `fetchVersionFromNpmTag()`
-- ✅ **Comprehensive Test Suite**: 29 new tests covering update flags functionality
-
-#### Technical Improvements
-- **NPM Integration**: Direct npm registry access for version queries
-- **Version Comparison Logic**: Robust semantic version comparison with prerelease handling
-- **Error Handling**: Comprehensive error messages with migration instructions
-- **User Experience**: One-command channel switching without data loss
-- **Backward Compatibility**: Zero breaking changes, existing workflows preserved
-
-### Version 4.5.1 - UI Quality Gate Fixes & Layout Improvements
-**Release Date**: 2025-12-08
-
-#### UI Fixes & Improvements
-- ✅ **Analytics UI Enhancements**: Standardized colors, fixed truncated model names, and ensured color consistency in the analytics dashboard.
-- ✅ **Auto-formatting**: 31 UI files auto-formatted for consistent styling.
-- ✅ **Fast Refresh Exports**: Resolved `react-refresh/only-export-components` by extracting `buttonVariants`, `useSidebar`, and `useWebSocketContext` to separate files.
-- ✅ **React Hooks Issues**: Fixed `react-hooks/purity` (`Math.random()` in `useMemo` for `sidebar.tsx`) and `react-hooks/set-state-in-effect` (`use-theme.ts`, `settings.tsx`).
-- ✅ **useWebSocket Hook Restructure**: Addressed `react-hooks/immutability` errors and dependency array warnings in `use-websocket.ts`.
-- ✅ **TypeScript Strict Mode**: Implemented null-check for `document.getElementById('root')` in `src/main.tsx` for strict mode compliance.
-- ✅ **Duplicate Directory Removal**: Cleaned up extraneous `ui/@/` directory.
-- ✅ **CLIProxy Card Padding**: Removed excessive padding from CLIProxy cards for better visual integration.
-- ✅ **CLIProxy Dashboard Layout**: Improved overall layout and styling of the CLIProxy dashboard.
-- ✅ **Dropdown Styling**: Refined dropdown component styling.
-- ✅ **Model Usage Card**: Corrected icon display and refined donut chart styling.
-
-#### Technical Improvements
-- **Improved UI Responsiveness**: Adjustments ensure better display across various screen sizes.
-- **Enhanced User Experience**: Minor visual tweaks lead to a more polished and intuitive interface.
-
-#### Validation Results
-- **UI Rendering**: ✅ All UI components render correctly after adjustments and fixes.
-- **Functional Impact**: ✅ No regressions introduced, core functionality remains stable.
-- **Code Quality**: ✅ All ESLint and TypeScript quality gates passed after fixes.
-
-### Version 4.5.2 - Operational Hub Redesign - Phase 1 Complete
-**Release Date**: 2025-12-12
-
-#### New UI Components Created
-- ✅ **ProfileCard Component** (`src/components/profile-card.tsx`): Individual profile display with status indicators and action buttons
-- ✅ **ProfileDeck Component** (`src/components/profile-deck.tsx`): Container for managing multiple profile cards in a grid layout
-- ✅ **CommandBuilder Component** (`src/components/command-builder.tsx`): Interactive command construction interface with suggestions
-- ✅ **ValueMetrics Component** (`src/components/value-metrics.tsx`): Dashboard widget for displaying efficiency metrics and cost savings
-- ✅ **HubFooter Component** (`src/components/hub-footer.tsx`): Utility footer with navigation links and version information
-
-#### Component Features
-- **Profile Management**: Visual profile switching with active/inactive states
-- **Command Interface**: Interactive command builder with copy/run actions
-- **Metrics Visualization**: Real-time display of cost savings and efficiency statistics
-- **Responsive Design**: Mobile-friendly layout with Tailwind CSS styling
-- **Accessibility**: WCAG-compliant components with proper ARIA labels
-
-#### Technical Implementation
-- **TypeScript**: Full type safety with zero `any` types
-- **React 19**: Latest React features with hooks and concurrent rendering
-- **shadcn/ui**: Consistent design system with Radix UI primitives
-- **Modular Architecture**: Reusable components for enhanced maintainability
-
-#### Code Review Notes
-- Identified console.log statements for removal in Phase 2
-- Noted hardcoded version string in HubFooter requiring dynamic version
-- Button handlers need implementation or disabling
-- Error boundary wrappers recommended for production
-
-#### Validation Results
-- **Component Compilation**: ✅ All components compile without errors
-- **TypeScript Validation**: ✅ Zero type errors with strict mode
-- **ESLint Compliance**: ✅ No violations detected
-- **Manual Testing**: ✅ Components render as expected in isolation
-
----
-
-#### Testing Infrastructure
-- **Version Comparison Tests**: 25 unit tests for all semantic version scenarios
-- **Flag Parsing Tests**: 4 integration tests for update command flags
-- **Edge Case Coverage**: Invalid versions, large numbers, case sensitivity
-- **Downgrade Detection**: Proper warnings for beta channel downgrades
-- **Test Framework**: Mocha with assert module, minimal mocking approach
-
-### Version 4.4.0 - TypeScript Conversion Complete
-**Release Date**: 2025-11-25
-
-#### Major Features
-- ✅ **Complete TypeScript Conversion**: All 31 source files migrated to TypeScript
-- ✅ **Zero `any` Types**: 100% type coverage with comprehensive type definitions
-- ✅ **Enhanced Build Pipeline**: Automated compilation with source maps and declarations
-- ✅ **Shared Settings Architecture**: Unified `settings.json` across all profiles
-- ✅ **Plugin Support**: Enhanced shared directory structure
-- ✅ **ESLint Strictness Upgrade**: Phase 01 completed (2025-11-27) - 3 rules upgraded to error level, 0 violations found
-- ✅ **CCS Architecture Refactoring**: Phase 02 completed (2025-11-27) - 44.6% file size reduction, 9 modular components created
-
-#### Technical Improvements
-- **Type Safety**: Compile-time error detection eliminates entire bug categories
-- **Developer Experience**: Full IDE support with IntelliSense and navigation
-- **Maintainability**: Self-documenting code with comprehensive type definitions
-- **Cross-Platform**: Typed abstractions ensure consistent behavior
-
-#### Breaking Changes
-- **None**: Fully backward compatible with existing configurations
-- **Migration**: Seamless upgrade path with zero user impact
-
-#### Dependencies
-- **TypeScript 5.3**: Updated to latest stable version
-- **Enhanced Build Process**: Automated compilation and type checking
-- **Development Tools**: Improved development tooling and debugging
-
-### Recent Patch Updates
-- **v4.3.10**: Package manager cache clearing during updates
-- **v4.3.9**: Fixed missing `commands/ccs.md` symlink in npm install
-- **v4.3.8**: Resolved missing `~/.ccs/.claude/` directory creation
-- **v4.3.7**: Enhanced directory creation during npm install
-- **v4.3.6**: Added plugin support to shared directories
-- **v4.4.0**: TypeScript conversion complete, shared settings architecture
-
-## Testing Strategy
-
-### TypeScript Testing Infrastructure
-1. **Unit Testing with Type Safety**
- - Mocha tests with TypeScript compilation
- - Type-safe test utilities and fixtures
- - Mock implementations with interface compliance
- - Coverage reporting with type metrics
-
-2. **Integration Testing**
- - Cross-platform compatibility validation
- - End-to-end workflow testing with typed data
- - API contract testing with type validation
- - Performance testing with type-aware profiling
-
-3. **Type-Level Testing**
- - TypeScript compiler error verification
- - Type coverage measurement and reporting
- - API surface validation with type checking
- - Dependency graph analysis with type relationships
-
-## Quality Assurance
-
-### TypeScript Quality Metrics
-- **Type Coverage**: 100% (target: maintain zero any types)
-- **Compiler Strictness**: Maximum strict mode configuration
-- **Interface Compliance**: All exports properly typed
-- **Documentation Coverage**: Type definitions serve as documentation
-
-### Code Quality Standards
-- **ESLint Integration**: TypeScript-specific linting rules
-- **Pre-commit Hooks**: Automated type checking before commits
-- **CI/CD Pipeline**: Type validation in continuous integration
-- **Code Review**: Type safety as review requirement
-
-## Contributing to Roadmap
-
-### TypeScript Development Guidelines
-1. **Type-First Development**: Define types before implementation
-2. **Zero Any Policy**: Maintain 100% type coverage
-3. **Interface Documentation**: Comprehensive JSDoc with type examples
-4. **Backward Compatibility**: Evolve APIs without breaking changes
-
-### Feature Contribution Process
-1. **Type Design**: Submit type definitions for review
-2. **Implementation**: Type-safe implementation with comprehensive tests
-3. **Documentation**: Update type documentation and examples
-4. **Review**: Peer review focused on type safety and API design
-
-## Community and Ecosystem
-
-### TypeScript Community Engagement
-- **TypeScript Best Practices**: Share learnings with TypeScript community
-- **Open Source Contribution**: Contribute to TypeScript tooling ecosystem
-- **Knowledge Sharing**: Document TypeScript migration experience
-- **Community Support**: Help other projects with TypeScript adoption
-
-## Future Roadmap
-
-### Phase 1: UI Enhancements & Mobile Support (v4.6.0 - v4.7.0)
-**Timeline**: 2-3 months
-
-**Priorities**:
-1. **Dashboard Improvements**
- - Sidebar redesign with modern UX patterns
- - Enhanced data visualizations and charts
- - Drag-and-drop profile reordering
- - Customizable dashboard widgets
-
-2. **Mobile Responsiveness**
- - Touch-friendly interface
- - Progressive Web App (PWA) support
- - Offline mode for basic functionality
- - Mobile-specific shortcuts
-
-3. **Real-time Collaboration**
- - Multi-user dashboard access
- - Live profile sharing
- - Collaborative delegation sessions
- - Real-time activity feeds
-
-### Phase 2: Advanced Features (v4.8.0 - v4.9.0)
-**Timeline**: 3-4 months
-
-**Priorities**:
-1. **AI-Powered Features**
- - Automatic model selection based on task type
- - Intelligent delegation routing
- - Cost optimization suggestions
- - Performance analytics
-
-2. **Enterprise Features**
- - Team profile management
- - Usage analytics dashboard
- - Role-based access control
- - Audit logging
-
-3. **Plugin System**
- - Extensible architecture for custom providers
- - Community plugin marketplace
- - Plugin development SDK
- - Version compatibility management
-
-### Phase 3: Next Generation Architecture (v5.0+)
-**Timeline**: 6+ months
-
-**Vision**:
-1. **Microservices Architecture**
- - Decoupled services for scalability
- - Container-based deployment
- - API-first design
- - Cloud-native features
-
-2. **Advanced AI Integration**
- - Multi-modal AI support (text, image, audio)
- - Custom AI model training
- - Advanced reasoning capabilities
- - Workflow automation
-
-3. **Ecosystem Expansion**
- - VS Code extension
- - IDE integrations
- - CI/CD plugins
- - Developer toolchain integration
-
-## Technical Debt & Improvements
-
-### Immediate Priorities (v4.5.1)
-1. **Performance Optimization**
- - Bundle size reduction
- - Lazy loading implementation
- - Memory leak fixes
- - Caching strategies
-
-2. **Testing Enhancement**
- - E2E test suite for UI
- - Visual regression testing
- - Performance testing
- - Accessibility testing
-
-3. **Documentation**
- - API documentation generation
- - Interactive tutorials
- - Video guides
- - Community resources
-
-### Medium-term Goals (v4.6.0)
-1. **Security Hardening**
- - Security audit
- - Penetration testing
- - Dependency vulnerability scanning
- - Secure development practices
-
-2. **Internationalization**
- - Multi-language support
- - Localization infrastructure
- - Cultural adaptations
- - Global deployment
-
-3. **Analytics & Monitoring**
- - Usage analytics
- - Performance monitoring
- - Error tracking
- - User behavior insights
-
----
-
-**Document Status**: Living document, updated with each major release
-**Last Updated**: 2025-12-12 (Operational Hub Redesign - Phase 1)
-**Next Update**: v4.5.3 Operational Hub Redesign - Phase 2 Integration
-**Maintainer**: CCS Development Team
\ No newline at end of file
diff --git a/docs/session-sharing-technical-analysis.md b/docs/session-sharing-technical-analysis.md
deleted file mode 100644
index 2798f4e8..00000000
--- a/docs/session-sharing-technical-analysis.md
+++ /dev/null
@@ -1,239 +0,0 @@
-# Why Session Sharing Fails Across Profiles: Technical Analysis
-
-## Overview
-
-Claude CLI sessions cannot be directly shared across CCS profiles due to authentication architecture constraints. When a user starts a thread on a "work" profile and the account expires, resuming from a "personal" profile fails despite the local session file being present.
-
-**Core Problem**: Thread IDs stored locally, but API-side validation requires matching OAuth credentials. Session "ownership" bound to account that created it, preventing cross-account continuation.
-
-## Technical Architecture: How Claude CLI Sessions Work
-
-### Session Storage Structure
-
-```
-~/.claude/
-├── .credentials.json # OAuth tokens (access, refresh, expiry)
-├── projects/
-│ └──
-
-# CCS - Claude Code Switch
-
-
-
-### Trình quản lý profile AI đa năng cho Claude Code
-
-**Chuyển đổi giữa nhiều tài khoản Claude, kết nối bất kỳ API tương thích Anthropic, và sử dụng OAuth providers (Gemini, Codex, Antigravity) ngay lập tức.**
-Ngừng bị rate limits. Làm việc liên tục với vô số profiles.
-
-
- -[](LICENSE) -[]() -[](https://www.npmjs.com/package/@kaitranntt/ccs) -[](https://claudekit.cc?ref=HMNKXOHN) - -**Languages**: [English](../../README.md) · [Tiếng Việt](README.md) · [日本語](../ja/README.md) - -
-
-- -[](LICENSE) -[]() -[](https://www.npmjs.com/package/@kaitranntt/ccs) -[](https://claudekit.cc?ref=HMNKXOHN) - -**Languages**: [English](../../README.md) · [Tiếng Việt](README.md) · [日本語](../ja/README.md) - -
- -## Bắt Đầu Nhanh - -### Cài Đặt - -**npm Package (Được khuyến nghị)** - -**macOS / Linux / Windows** -```bash -npm install -g @kaitranntt/ccs -``` - -**Tất cả các trình quản lý package chính đều được hỗ trợ:** - -```bash -# yarn -yarn global add @kaitranntt/ccs - -# pnpm (ít hơn 70% dung lượng đĩa) -pnpm add -g @kaitranntt/ccs - -# bun (nhanh hơn 30x) -bun add -g @kaitranntt/ccs -``` - -
-
- -> [!WARNING] -> **Các trình cài đặt này đã lỗi thời và sẽ bị xóa trong phiên bản tương lai.** -> Hiện tại chúng tự động chuyển hướng đến cài đặt npm. Vui lòng sử dụng npm trực tiếp. - -**macOS / Linux** -```bash -curl -fsSL ccs.kaitran.ca/install | bash -``` - -**Windows PowerShell** -```powershell -irm ccs.kaitran.ca/install | iex -``` - -**Lưu ý**: Script hiển thị cảnh báo lỗi thời và tự động chạy cài đặt npm nếu Node.js khả dụng. - -
-
-[!] LỖI THỜI: Trình cài đặt shell gốc (Cũ)
- -- -> [!WARNING] -> **Các trình cài đặt này đã lỗi thời và sẽ bị xóa trong phiên bản tương lai.** -> Hiện tại chúng tự động chuyển hướng đến cài đặt npm. Vui lòng sử dụng npm trực tiếp. - -**macOS / Linux** -```bash -curl -fsSL ccs.kaitran.ca/install | bash -``` - -**Windows PowerShell** -```powershell -irm ccs.kaitran.ca/install | iex -``` - -**Lưu ý**: Script hiển thị cảnh báo lỗi thời và tự động chạy cài đặt npm nếu Node.js khả dụng. - -
- -### Cấu Hình (Tự Tạo) - -**CCS tự động tạo cấu hình trong quá trình cài đặt** (thông qua script postinstall của npm). - -**~/.ccs/config.json**: -```json -{ - "profiles": { - "glm": "~/.ccs/glm.settings.json", - "glmt": "~/.ccs/glmt.settings.json", - "kimi": "~/.ccs/kimi.settings.json", - "default": "~/.claude/settings.json" - } -} -``` - -
-
-
-
- -Nếu Claude CLI được cài đặt ở vị trí không chuẩn (ổ D, thư mục tùy chỉnh), đặt `CCS_CLAUDE_PATH`: - -```bash -# Unix/Linux/macOS -export CCS_CLAUDE_PATH="/path/to/claude" - -# Windows PowerShell -$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe" -``` - -**Xem thêm**: [Hướng dẫn Khắc phục Sự cố](./docs/en/troubleshooting.md#claude-cli-in-non-standard-location) để biết chi tiết cài đặt. - -
-
-Custom Claude CLI Path
-
-- -Nếu Claude CLI được cài đặt ở vị trí không chuẩn (ổ D, thư mục tùy chỉnh), đặt `CCS_CLAUDE_PATH`: - -```bash -# Unix/Linux/macOS -export CCS_CLAUDE_PATH="/path/to/claude" - -# Windows PowerShell -$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe" -``` - -**Xem thêm**: [Hướng dẫn Khắc phục Sự cố](./docs/en/troubleshooting.md#claude-cli-in-non-standard-location) để biết chi tiết cài đặt. - -
-
-
-
- -**Người dùng Windows**: Bật Chế độ Nhà phát triển để có symlink thực sự (hiệu suất tốt hơn, đồng bộ hóa tức thì): - -1. Mở **Settings** → **Privacy & Security** → **For developers** -2. Bật **Developer Mode** -3. Cài đặt lại CCS: `npm install -g @kaitranntt/ccs` - -**Cảnh báo**: Nếu không có Chế độ Nhà phát triển, CCS tự động chuyển sang sao chép thư mục (hoạt động nhưng không đồng bộ tức thì trên các profile). - -
-
-Windows Symlink Support (Developer Mode)
-
-- -**Người dùng Windows**: Bật Chế độ Nhà phát triển để có symlink thực sự (hiệu suất tốt hơn, đồng bộ hóa tức thì): - -1. Mở **Settings** → **Privacy & Security** → **For developers** -2. Bật **Developer Mode** -3. Cài đặt lại CCS: `npm install -g @kaitranntt/ccs` - -**Cảnh báo**: Nếu không có Chế độ Nhà phát triển, CCS tự động chuyển sang sao chép thư mục (hoạt động nhưng không đồng bộ tức thì trên các profile). - -
- -### Lần Chuyển Đổi Đầu Tiên - -> [!IMPORTANT] -> **Trước khi dùng các mô hình thay thế, cập nhật API keys trong file settings:** -> -> - **GLM**: Chỉnh sửa `~/.ccs/glm.settings.json` và thêm Z.AI Coding Plan API Key của bạn -> - **GLMT**: Chỉnh sửa `~/.ccs/glmt.settings.json` và thêm Z.AI Coding Plan API Key của bạn -> - **Kimi**: Chỉnh sửa `~/.ccs/kimi.settings.json` và thêm Kimi API key của bạn - -
- -**Parallel Workflow: Planning + Execution** - -```bash -# Terminal 1 - Planning (Claude Sonnet) -ccs "Plan a REST API with authentication and rate limiting" - -# Terminal 2 - Execution (GLM, cost-optimized) -ccs glm "Implement the user authentication endpoints from the plan" -``` - -
-
- -```bash -# Kimi - Stable thinking support -ccs kimi "Design a caching strategy with trade-off analysis" - -# GLMT - Experimental (see full disclaimer below) -ccs glmt "Debug complex algorithm with reasoning steps" -``` - -**Lưu ý:** GLMT là thử nghiệm và không ổn định. Xem phần [GLM with Thinking (GLMT)](#glm-with-thinking-glmt) dưới đây để biết chi tiết. - -
-
-Thinking Models (Kimi & GLMT)
- -- -```bash -# Kimi - Stable thinking support -ccs kimi "Design a caching strategy with trade-off analysis" - -# GLMT - Experimental (see full disclaimer below) -ccs glmt "Debug complex algorithm with reasoning steps" -``` - -**Lưu ý:** GLMT là thử nghiệm và không ổn định. Xem phần [GLM with Thinking (GLMT)](#glm-with-thinking-glmt) dưới đây để biết chi tiết. - -
- -## The Daily Developer Pain Point - -
-
-### **DỪNG việc chuyển đổi. BẮT ĐẦU điều phối.**
-
-**Giới hạn phiên không nên phá hỏng trạng thái dòng chảy của bạn.**
-
-
-Bạn đang sâu trong triển khai. Ngữ cảnh đã tải. Giải pháp đang kết tinh.-Sau đó: 🔴 _"Bạn đã đạt đến giới hạn sử dụng."_ - -**Động lực mất đi. Ngữ cảnh mất. Năng suất sụp đổ.** - -## **Giải pháp: Quy trình công việc song song** - -
-
-
-❌ CÁCH CŨ: Chuyển đổi khi bạn đạt đến giới hạn (Phản ứng)
- -### Quy trình làm việc hiện tại của bạn: -- **2pm:** Xây dựng tính năng, trong vùng -- **3pm:** 🔴 Đạt giới hạn sử dụng -- **3:05pm:** Dừng công việc, chỉnh sửa `~/.claude/settings.json` -- **3:15pm:** Chuyển tài khoản, mất ngữ cảnh -- **3:30pm:** Cố gắng quay lại trạng thái dòng chảy -- **4pm:** Cuối cùng cũng năng suất trở lại - -- **Kết quả:** Mất 1 giờ, động lực bị phá hủy, sự thất vọng tăng lên - -
-
-
-## Chọn con đường của bạn
-
-✨ CÁCH MỚI: Chạy song song ngay từ đầu (Chủ động) - ĐƯỢC KHUYÊN NGHỊ
- -### Quy trình làm việc mới của bạn: -- **2pm:** **Terminal 1:** `ccs "Lập kế hoạch kiến trúc API"` → Tư duy chiến lược (Claude Pro) -- **2pm:** **Terminal 2:** `ccs glm "Triển khai các điểm cuối API"` → Thực thi mã (GLM) -- **3pm:** Vẫn đang giao hàng, không có gián đoạn -- **4pm:** Đạt trạng thái dòng chảy, năng suất tăng vọt -- **5pm:** Tính năng đã giao hàng, ngữ cảnh được duy trì - -- **Kết quả:** Không có thời gian chết, năng suất liên tục, ít thất vọng hơn - -### 💰 **Giá trị đề xuất:** -- **Thiết lập:** Claude Pro hiện tại của bạn + GLM Lite (add-on hiệu quả về chi phí) -- **Giá trị:** Tiết kiệm 1 giờ/ngày × 20 ngày làm việc = 20 giờ/tháng được phục hồi -- **ROI:** Thời gian phát triển của bạn có giá trị hơn chi phí thiết lập -- **Thực tế:** Giao hàng nhanh hơn chi phí vận hành - -
-
-
-Tập trung vào ngân sách: Chỉ GLM
- -- **Tốt nhất cho:** Phát triển tiết kiệm chi phí, tạo mã cơ bản -- **Sử dụng:** Chỉ sử dụng `ccs glm` trực tiếp để được trợ giúp AI hiệu quả về chi phí -- **Thực tế:** Không có quyền truy cập Claude, nhưng có khả năng cho nhiều nhiệm vụ mã hóa -- **Thiết lập:** Chỉ cần API key GLM, rất phải chăng - -
-
-
-✨ Được khuyên nghị cho phát triển hàng ngày: 1 Claude Pro + 1 GLM Lite
- -- **Tốt nhất cho:** Giao hàng mã hàng ngày, công việc phát triển nghiêm túc -- **Sử dụng:** `ccs` để lập kế hoạch + `ccs glm` để thực thi (quy trình công việc song song) -- **Thực tế:** Cân bằng hoàn hảo giữa khả năng và chi phí cho hầu hết các nhà phát triển -- **Giá trị:** Không bao giờ đạt đến giới hạn phiên, năng suất liên tục - -
-
-
-Power User: Nhiều Claude Pro + GLM Pro
- -- **Tốt nhất cho:** Nhiều công việc, dự án đồng thời, solo dev -- **Mở khóa:** Không bao giờ cạn kiệt giới hạn phiên hoặc hàng tuần -- **Quy trình làm việc:** 3+ terminal chạy các nhiệm vụ chuyên biệt đồng thời - -
-
-
----
-
-## Why CCS Instead of Manual Switching?
-
-Tập trung vào quyền riêng tư: Cách ly Công việc/Cá nhân
- -- **Khi cần:** Cách ly nghiêm ngặt ngữ cảnh AI công việc và cá nhân -- **Thiết lập:** `ccs auth create work` + `ccs auth create personal` -- **Lưu ý:** Tính năng nâng cao - hầu hết người dùng không cần điều này - -
-
-**CCS không phải về "chuyển đổi khi bạn đạt đến giới hạn lúc 3pm."**
-
-## **Nó về việc chạy song song ngay từ đầu.**
-
-
-
-### Sự khác biệt cốt lõi
-
-| **Chuyển đổi thủ công** | **Điều phối CCS** |
-|:---|:---|
-| 🔴 Đạt giới hạn → Dừng công việc → Chỉnh sửa tệp cấu hình → Khởi động lại | ✅ Nhiều terminal chạy các mô hình khác nhau ngay từ đầu |
-| 😰 Mất ngữ cảnh và gián đoạn trạng thái dòng chảy | 😌 Năng suất liên tục với ngữ cảnh được bảo toàn |
-| 📝 Xử lý nhiệm vụ tuần tự | ⚡ Quy trình công việc song song (lập kế hoạch + thực thi đồng thời) |
-| 🛠️ Giải quyết vấn đề phản ứng khi bị chặn | 🎯 Thiết kế quy trình công việc chủ động ngăn chặn chặn |
-
-### CCS mang lại cho bạn
-
-- **Không chuyển đổi ngữ cảnh:** Duy trì trạng thái dòng chảy của bạn mà không bị gián đoạn
-- **Năng suất song song:** Lập kế hoạch chiến lược trong một terminal, thực thi mã trong terminal khác
-- **Quản lý tài khoản tức thì:** Một lệnh chuyển đổi, không cần chỉnh sửa tệp cấu hình
-- **Cách ly công việc-cuộc sống:** Cách ly ngữ cảnh mà không cần đăng xuất
-- **Tính nhất quán đa nền tảng:** Trải nghiệm mượt mà tương tự trên macOS, Linux, Windows
-
-- -## Architecture - -### Profile Types - -**Settings-based**: GLM, GLMT, Kimi, default -- Uses `--settings` flag pointing to config files -- GLMT: Embedded proxy for thinking mode support - -**Account-based**: work, personal, team -- Uses `CLAUDE_CONFIG_DIR` for isolated instances -- Create with `ccs auth create
- -## Usage Examples - -### Basic Switching - -```bash -ccs # Claude subscription (default) -ccs glm # GLM (cost-optimized) -ccs kimi # Kimi (with thinking support) -``` - -### Multi-Account Setup - -```bash -# Create accounts -ccs auth create work -ccs auth create personal -``` - -**Run concurrently in separate terminals:** - -```bash -# Terminal 1 - Work -ccs work "implement feature" - -# Terminal 2 - Personal (concurrent) -ccs personal "review code" -``` - -### Help & Version - -```bash -ccs --version # Show version -ccs --help # Show all commands and options -``` - -
- -## GLM with Thinking (GLMT) - -> [!CAUTION] -> ### NOT PRODUCTION READY - EXPERIMENTAL FEATURE -> -> **GLMT is experimental and requires extensive debugging**: -> - Streaming and tool support still under active development -> - May experience unexpected errors, timeouts, or incomplete responses -> - Requires frequent debugging and manual intervention -> - **Not recommended for critical workflows or production use** -> -> **Alternative for GLM Thinking**: Consider going through the **CCR hustle** with the **Transformer of Bedolla** ([ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/)) for a more stable implementation. - -> [!IMPORTANT] -> GLMT requires npm installation (`npm install -g @kaitranntt/ccs`). Not available in native shell versions (requires Node.js HTTP server). - -
- -> [!NOTE] -> ### Acknowledgments: The Foundation That Made GLMT Possible -> -> **CCS's GLMT implementation owes its existence to the groundbreaking work of [@Bedolla](https://github.com/Bedolla)**, who created [ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/) - the **first integration** to bridge [Claude Code Router (CCR)](https://github.com/musistudio/claude-code-router) with Z.AI's reasoning capabilities. -> -> Before ZaiTransformer, no one had successfully integrated Z.AI's thinking mode with Claude Code's workflow. Bedolla's work wasn't just helpful - it was **foundational**. His implementation of request/response transformation architecture, thinking mode control mechanisms, and embedded proxy design directly inspired and enabled GLMT's design. -> -> **Without ZaiTransformer's pioneering work, GLMT wouldn't exist in its current form.** If you benefit from GLMT's thinking capabilities, please consider starring [ZaiTransformer](https://github.com/Bedolla/ZaiTransformer/) to support pioneering work in the Claude Code ecosystem. - -
- -
-
-
-
- -
-
-GLM vs GLMT Comparison
-
-- -
-
-| Feature | GLM (`ccs glm`) | GLMT (`ccs glmt`) |
-|:--------|:----------------|:------------------|
-| **Endpoint** | Anthropic-compatible | OpenAI-compatible |
-| **Thinking** | No | Experimental (`reasoning_content`) |
-| **Tool Support** | Basic | **Unstable (v3.5+)** |
-| **MCP Tools** | Limited | **Buggy (v3.5+)** |
-| **Streaming** | Stable | **Experimental (v3.4+)** |
-| **TTFB** | <500ms | <500ms (sometimes), 2-10s+ (often) |
-| **Use Case** | Reliable work | **Debugging experiments only** |
-
-
-
-- -
-
-
-
- -**GLMT attempts MCP tools and function calling:** - -- **Bidirectional Transformation**: Anthropic tools ↔ OpenAI format (unstable) -- **MCP Integration**: MCP tools sometimes execute (often output XML garbage) -- **Streaming Tool Calls**: Real-time tool calls (when not crashing) -- **Backward Compatible**: May break existing thinking support -- **Configuration Required**: Frequent manual debugging needed - -
-
-Tool Support (v3.5) - EXPERIMENTAL
-
-- -**GLMT attempts MCP tools and function calling:** - -- **Bidirectional Transformation**: Anthropic tools ↔ OpenAI format (unstable) -- **MCP Integration**: MCP tools sometimes execute (often output XML garbage) -- **Streaming Tool Calls**: Real-time tool calls (when not crashing) -- **Backward Compatible**: May break existing thinking support -- **Configuration Required**: Frequent manual debugging needed - -
-
-
-
- -**GLMT attempts real-time streaming** with incremental reasoning content delivery: - -- **Default**: Streaming enabled (TTFB <500ms when it works) -- **Auto-fallback**: Frequently switches to buffered mode due to errors -- **Thinking parameter**: Claude CLI `thinking` parameter sometimes works - - May ignore `thinking.type` and `budget_tokens` - - Precedence: CLI parameter > message tags > default (when not broken) - -**Status**: Z.AI (tested, tool calls frequently break, requires constant debugging) - -
-
-Streaming Support (v3.4) - OFTEN FAILS
-
-- -**GLMT attempts real-time streaming** with incremental reasoning content delivery: - -- **Default**: Streaming enabled (TTFB <500ms when it works) -- **Auto-fallback**: Frequently switches to buffered mode due to errors -- **Thinking parameter**: Claude CLI `thinking` parameter sometimes works - - May ignore `thinking.type` and `budget_tokens` - - Precedence: CLI parameter > message tags > default (when not broken) - -**Status**: Z.AI (tested, tool calls frequently break, requires constant debugging) - -
-
-
-
- -1. CCS spawns embedded HTTP proxy on localhost (if not crashing) -2. Proxy attempts to convert Anthropic format → OpenAI format (often fails) -3. Tries to transform Anthropic tools → OpenAI function calling format (buggy) -4. Forwards to Z.AI with reasoning parameters and tools (when not timing out) -5. Attempts to convert `reasoning_content` → thinking blocks (partial or broken) -6. Attempts to convert OpenAI `tool_calls` → Anthropic `tool_use` blocks (XML garbage common) -7. Thinking and tool calls sometimes appear in Claude Code UI (when not broken) - -
-
-How It Works (When It Works)
-
-- -1. CCS spawns embedded HTTP proxy on localhost (if not crashing) -2. Proxy attempts to convert Anthropic format → OpenAI format (often fails) -3. Tries to transform Anthropic tools → OpenAI function calling format (buggy) -4. Forwards to Z.AI with reasoning parameters and tools (when not timing out) -5. Attempts to convert `reasoning_content` → thinking blocks (partial or broken) -6. Attempts to convert OpenAI `tool_calls` → Anthropic `tool_use` blocks (XML garbage common) -7. Thinking and tool calls sometimes appear in Claude Code UI (when not broken) - -
-
-
-
- -**Control Tags**: -- `` - Enable/disable reasoning blocks (default: On)
-- `` - Control reasoning depth (deprecated - Z.AI only supports binary thinking)
-
-**Thinking Keywords** (inconsistent activation):
-- `think` - Sometimes enables reasoning (low effort)
-- `think hard` - Sometimes enables reasoning (medium effort)
-- `think harder` - Sometimes enables reasoning (high effort)
-- `ultrathink` - Attempts maximum reasoning depth (often breaks)
-
-
-
-Control Tags & Keywords
-
-- -**Control Tags**: -- `
-
-
-
- -**GLMT features** (all experimental): -- Forced English output enforcement (sometimes works) -- Random thinking mode activation (unpredictable) -- Attempted streaming with frequent fallback to buffered mode - -**General**: -- `CCS_DEBUG_LOG=1` - Enable debug file logging -- `CCS_CLAUDE_PATH=/path/to/claude` - Custom Claude CLI path - -
-
-Environment Variables
-
-- -**GLMT features** (all experimental): -- Forced English output enforcement (sometimes works) -- Random thinking mode activation (unpredictable) -- Attempted streaming with frequent fallback to buffered mode - -**General**: -- `CCS_DEBUG_LOG=1` - Enable debug file logging -- `CCS_CLAUDE_PATH=/path/to/claude` - Custom Claude CLI path - -
-
-
-
- -```bash -# Edit GLMT settings -nano ~/.ccs/glmt.settings.json -``` - -Set Z.AI API key (requires coding plan): - -```json -{ - "env": { - "ANTHROPIC_AUTH_TOKEN": "your-z-ai-api-key" - } -} -``` - -
-
-API Key Setup
-
-- -```bash -# Edit GLMT settings -nano ~/.ccs/glmt.settings.json -``` - -Set Z.AI API key (requires coding plan): - -```json -{ - "env": { - "ANTHROPIC_AUTH_TOKEN": "your-z-ai-api-key" - } -} -``` - -
-
-
-
- -**v3.4 Protection Limits**: - -| Limit | Value | Purpose | -|:------|:------|:--------| -| **SSE buffer** | 1MB max per event | Prevent buffer overflow | -| **Content buffer** | 10MB max per block | Limit thinking/text blocks | -| **Content blocks** | 100 max per message | Prevent DoS attacks | -| **Request timeout** | 120s | Both streaming and buffered | - -
-
-Security Limits (DoS Protection)
-
-- -**v3.4 Protection Limits**: - -| Limit | Value | Purpose | -|:------|:------|:--------| -| **SSE buffer** | 1MB max per event | Prevent buffer overflow | -| **Content buffer** | 10MB max per block | Limit thinking/text blocks | -| **Content blocks** | 100 max per message | Prevent DoS attacks | -| **Request timeout** | 120s | Both streaming and buffered | - -
-
-
-
- -**Enable verbose logging**: -```bash -ccs glmt --verbose "your prompt" -``` - -**Enable debug file logging**: -```bash -export CCS_DEBUG_LOG=1 -ccs glmt --verbose "your prompt" -# Logs: ~/.ccs/logs/ -``` - -**GLMT debugging**: -```bash -# Verbose logging shows streaming status and reasoning details -ccs glmt --verbose "test" -``` - -**Check reasoning content**: -```bash -cat ~/.ccs/logs/*response-openai.json | jq '.choices[0].message.reasoning_content' -``` - -**Troubleshooting**: -- **If absent**: Z.AI API issue (verify key, account status) -- **If present**: Transformation issue (check `response-anthropic.json`) - -
-
-Debugging
-
-- -**Enable verbose logging**: -```bash -ccs glmt --verbose "your prompt" -``` - -**Enable debug file logging**: -```bash -export CCS_DEBUG_LOG=1 -ccs glmt --verbose "your prompt" -# Logs: ~/.ccs/logs/ -``` - -**GLMT debugging**: -```bash -# Verbose logging shows streaming status and reasoning details -ccs glmt --verbose "test" -``` - -**Check reasoning content**: -```bash -cat ~/.ccs/logs/*response-openai.json | jq '.choices[0].message.reasoning_content' -``` - -**Troubleshooting**: -- **If absent**: Z.AI API issue (verify key, account status) -- **If present**: Transformation issue (check `response-anthropic.json`) - -
- -## Uninstall - -
-
-
-
- -```bash -# npm -npm uninstall -g @kaitranntt/ccs - -# yarn -yarn global remove @kaitranntt/ccs - -# pnpm -pnpm remove -g @kaitranntt/ccs - -# bun -bun remove -g @kaitranntt/ccs -``` - -
-
-Package Managers
-
-- -```bash -# npm -npm uninstall -g @kaitranntt/ccs - -# yarn -yarn global remove @kaitranntt/ccs - -# pnpm -pnpm remove -g @kaitranntt/ccs - -# bun -bun remove -g @kaitranntt/ccs -``` - -
-
-
-
- -```bash -# macOS / Linux -curl -fsSL ccs.kaitran.ca/uninstall | bash - -# Windows PowerShell -irm ccs.kaitran.ca/uninstall | iex -``` - -
-
-Official Uninstaller
-
-- -```bash -# macOS / Linux -curl -fsSL ccs.kaitran.ca/uninstall | bash - -# Windows PowerShell -irm ccs.kaitran.ca/uninstall | iex -``` - -
- -## 🎯 Philosophy - -- **YAGNI**: No features "just in case" -- **KISS**: Simple bash, no complexity -- **DRY**: One source of truth (config) - -## 📖 Documentation - -**Complete documentation in [docs/](./docs/)**: -- [Installation Guide](./docs/en/installation.md) -- [Configuration](./docs/en/configuration.md) -- [Usage Examples](./docs/en/usage.md) -- [System Architecture](./docs/system-architecture.md) -- [GLMT Control Mechanisms](./docs/glmt-controls.md) -- [Troubleshooting](./docs/en/troubleshooting.md) -- [Contributing](./CONTRIBUTING.md) - -## 🤝 Contributing - -We welcome contributions! Please see our [Contributing Guide](./CONTRIBUTING.md) for details. - -## Star History - -
-
-
-
-
-
-
-## License
-
-CCS is licensed under the [MIT License](LICENSE).
-
-
-
-**Made with ❤️ for developers who hit rate limits too often**
-
-[⭐ Star this repo](https://github.com/kaitranntt/ccs) | [🐛 Report issues](https://github.com/kaitranntt/ccs/issues) | [📖 Read docs](./docs/en/)
-
-
\ No newline at end of file
diff --git a/docs/vi/configuration.vi.md b/docs/vi/configuration.vi.md
deleted file mode 100644
index f272bdd9..00000000
--- a/docs/vi/configuration.vi.md
+++ /dev/null
@@ -1,164 +0,0 @@
-# Hướng Dẫn Cấu Hình CCS
-
-## Cấu Hình Tự Động
-
-Installer tự động tạo config và mẫu profile trong quá trình cài đặt:
-
-**macOS / Linux**: `~/.ccs/config.json`
-
-**Windows**: `%USERPROFILE%\.ccs\config.json`
-
-## Định Dạng Cấu Hình
-
-### Cài Đặt Cơ Bản
-
-```json
-{
- "profiles": {
- "glm": "~/.ccs/glm.settings.json",
- "default": "~/.claude/settings.json"
- }
-}
-```
-
-### Cài Đặt Nâng Cao (Nhiều Profile)
-
-```json
-{
- "profiles": {
- "glm": "~/.ccs/glm.settings.json",
- "haiku": "~/.ccs/haiku.settings.json",
- "custom": "~/.ccs/custom.settings.json",
- "default": "~/.claude/settings.json"
- }
-}
-```
-
-## Cấu Hình Profile
-
-### Ví Dụ Profile GLM
-
-**Vị trí**: `~/.ccs/glm.settings.json`
-
-```json
-{
- "env": {
- "ANTHROPIC_BASE_URL": "https://api.z.ai/api/anthropic",
- "ANTHROPIC_AUTH_TOKEN": "your_glm_api_key",
- "ANTHROPIC_MODEL": "glm-4.6",
- "ANTHROPIC_DEFAULT_OPUS_MODEL": "glm-4.6",
- "ANTHROPIC_DEFAULT_SONNET_MODEL": "glm-4.6",
- "ANTHROPIC_DEFAULT_HAIKU_MODEL": "glm-4.6"
- }
-}
-```
-
-### Profile Claude (Mặc Định)
-
-- Sử dụng `~/.claude/settings.json` (config Claude CLI hiện tại của bạn)
-- CCS không bao giờ sửa file này (tiếp cận không xâm phạm)
-
-## Cách Hoạt Động Cấu Hình
-
-1. CCS đọc tên profile từ dòng lệnh (mặc định là "default")
-2. Tìm đường dẫn file settings trong `~/.ccs/config.json`
-3. Thực thi `claude --settings