aidlc-mcp-server

# AIDLC MCP Server AI Development Life Cycle workflow guidance server, built on [FastMCP](https://github.com/jlowin/fastmcp). Dynamically loads workflow instructions from the bundled `aidlc-rules/` directory and guides LLM agents through a structured software development process. All project state lives in the user's workspace — the server is stateless. ## Quick Start ### Option 1: uvx (recommended for end users) No installation needed — `uvx` downloads and runs the server in an isolated environment: ```bash uvx aidlc-mcp-server ``` ### Option 2: pip install ```bash pip install aidlc-mcp-server aidlc-mcp-server ``` ### From source (development) ```bash git clone <this-monorepo> cd aidlc-mcp-server pip install -e ".[dev]" ./scripts/sync-aidlc-rules.sh # Copy rules from monorepo root aidlc-mcp-server ``` Workflow rules are bundled inside the package, so both `uvx` and `pip install` work out of the box. You can override the rules directory at runtime with `--workflow-dir /path/to/aidlc-rules` or the `AIDLC_WORKFLOW_DIR` environment variable. ## MCP Client Configuration ### Using uvx (no install required) ```json { "mcpServers": { "aidlc": { "command": "uvx", "args": ["aidlc-mcp-server"] } } } ``` ### Using pip install ```json { "mcpServers": { "aidlc": { "command": "aidlc-mcp-server", "args": [] } } } ``` Rules are bundled inside the package — no `cwd` or environment variables needed. ## Tools | Tool | Purpose | |---------------------------|----------------------------------------------| | `aidlc_start_project` | Create a new project with operational mode | | `aidlc_get_guidance` | Load workflow/stage guidance dynamically | | `aidlc_complete_stage` | Save deliverable, advance to next stage | | `aidlc_list_projects` | List all projects in workspace | | `aidlc_log` | Append to project audit …