integration.md

Integration Guide

PRD Pilot exposes a thin integration surface so external tools can reuse the same workflow as the web app.

Shared workflow

1. Structure the requirement.
2. Generate PRD and/or demo.
3. Run consistency checks.
4. Apply targeted iteration.

1) Cursor

If your Cursor configuration supports an MCP server definition, point it to the local MCP entrypoint from the repository root:

{
  "mcpServers": {
    "prd-pilot": {
      "command": "python",
      "args": ["mcp/mcp_service.py"],
      "cwd": "/path/to/prd-pilot-repo"
    }
  }
}

If you prefer a networked transport, start the server with:

python mcp/mcp_service.py --transport streamable-http --host 127.0.0.1 --port 8765

2) Cherry Studio

Cherry Studio can connect to the same MCP server using the same tool list.

Recommended local command from the repository root:

python mcp/mcp_service.py

If your client only accepts a URL-style transport, use streamable HTTP mode and point the client to http://127.0.0.1:8765.

3) Claude Code

This repository already includes a Claude Code skill at:

.claude/skills/prd-pilot/SKILL.md

Use it when you want Claude Code to follow the PRD Pilot workflow directly:

• structure requirement
• generate PRD / demo
• check consistency
• targeted iteration

Minimal backend startup

cd prd-pilot/backend
pip install -r requirements.txt
copy .env.example .env
python main.py

Tool mapping

• structure_requirement
• generate_prd
• generate_demo
• check_consistency
• iterate_prd
• iterate_demo