Skip to content

README.md

Latest commit

 

History

History
200 lines (141 loc) · 5.76 KB

README.md

File metadata and controls

200 lines (141 loc) · 5.76 KB

copilot-flow

Plugin Version License

繁體中文

AI collaboration workflow plugin - Implements automated collaborative development process between Claude and Copilot through Copilot MCP

✨ Features

  • 🔄 Automated 5-stage development workflow (Requirements Analysis → Architecture Design → Implementation → Review → Delivery)
  • 🤖 Smart model selection (Automatically selects the most suitable Copilot model based on task type)
  • 👀 Preview mode (Shows plan before execution, requires confirmation)
  • 🎯 Responsibility separation (Copilot provides suggestions, Claude executes code changes)
  • 🛠️ Error handling (Clear prompts when MCP is unavailable)
  • 📊 State management (Supports interruption recovery)
  • 💡 Optional memory (Recommend using claude-mem plugin for advanced session tracking)

🔧 Prerequisites

Required

  1. Claude Code CLI

    npm install -g @anthropic-ai/claude-code

  2. Copilot MCP Server (⚠️ Must install before plugin)

    npm install -g @aykahshi/copilot-mcp-server

  3. Copilot CLI Authentication

    npm install -g @github/copilot
copilot /login

🚀 Installation

Step 1: Install Copilot MCP Server (Required)

⚠️ IMPORTANT: You must install this globally first!

npm install -g @aykahshi/copilot-mcp-server

Verify installation:

copilot-mcp-server --version

Step 2: Install Plugin

# Add plugin marketplace
/plugin marketplace add Poorgramer-Zack/copilot-mcp-tool

# Install plugin
/plugin install copilot-flow

Usage

Execute Full Workflow

請 Copilot 協助我實現用戶認證功能

Stage-by-Stage Execution

/copilot-flow:analyze   # Requirements analysis
/copilot-flow:design    # Architecture design
/copilot-flow:implement # Implementation
/copilot-flow:review    # Review
/copilot-flow:deliver   # Delivery

Shortcuts

c-flow:analyze   # Analyze requirements
c-flow:design    # Design architecture
c-flow:implement # Implement code
c-flow:review    # Review code
c-flow:deliver   # Deliver results

Model Selection Rules

  • Architecture Design and Code Issues → claude-sonnet-4.6
  • Quick Code QA → claude-haiku-4.5
  • Non-code Quick QA → gpt-5-mini
  • High-Difficulty Complex Tasks → gpt-5.2-codex

Workflow

  1. Preview Mode: Claude proposes initial plan
  2. Execute after Confirmation: Enter Copilot interaction flow
  3. Automatic Context: Leverage Claude's native memory for conversation continuity
  4. Recovery Mechanism: Can recover interrupted workflows through session context

📝 Session Recording (Optional)

copilot-flow focuses on workflow automation and doesn't include built-in session recording. If you need advanced memory and context management, we recommend using:

💡 Recommended: claude-mem Plugin

The claude-mem plugin provides powerful memory capabilities:

  • Automatic Memory: Captures and stores conversation context
  • Semantic Search: Find relevant past conversations
  • Long-term Context: Maintain context across multiple sessions

Installation:

/plugin marketplace add thedotmack/claude-mem

/plugin install claude-mem

Benefits for copilot-flow:

  • Automatically remembers your workflow decisions
  • Recalls past implementations and patterns
  • Provides better context for multi-session projects
  • No manual session management needed

⚠️ Important Notes

  • 🎯 Responsibility Separation: Copilot only provides suggestions, does not directly modify code
  • 💻 Code Execution: All code modifications are executed by Claude based on Copilot suggestions
  • 🔗 MCP Dependency: Ensure Copilot MCP server is available
  • 💾 State Persistence: Workflow state saved in .claude/workflow-state.json
  • 📝 Memory: Use claude-mem plugin for advanced session recording (optional)
  • 🔄 Minimal Dependencies: Core functionality works out of the box

🛠️ Troubleshooting

Common Issues

Q: MCP Server status shows "failed"?

A: Make sure you have installed @aykahshi/copilot-mcp-server globally:

npm install -g @aykahshi/copilot-mcp-server

# Verify installation
copilot-mcp-server --version

Then restart Claude Desktop completely.

Q: Commands not showing in help?

A:

  1. Check if plugin is properly installed: /plugin list
  2. Verify MCP server is installed globally
  3. Restart Claude Desktop completely

Q: "copilot-mcp-server: command not found"?

A: The MCP server is not installed or not in PATH:

# Install globally
npm install -g @aykahshi/copilot-mcp-server

# On Windows, you may need to restart your terminal/Claude Desktop
# On macOS/Linux, ensure npm global bin is in PATH:
export PATH="$PATH:$(npm config get prefix)/bin"

Q: Workflow stuck?

A: Delete .claude/workflow-state.json to reset

Q: Need to check MCP connection?

A: Run the MCP server directly to test:

copilot-mcp-server
# Should start without errors
# Press Ctrl+C to exit

📚 Documentation

🤝 Contributing

Issues and Pull Requests are welcome!

📄 License

MIT License - See LICENSE file