MCPForge is a lightweight toolkit for scaffolding and validating Model Context Protocol servers locally.
This first slice focuses on two practical workflows:
mcpforge initcreates a minimal stdio MCP starter projectmcpforge checkruns a smoke test against a local server and validates the basic MCP handshake
MCP server development still has a lot of repeated setup work:
- creating a starter project structure
- wiring up a local stdio server loop
- checking whether the server starts and responds correctly
- documenting a runnable local workflow for the next contributor
MCPForge exists to make that first setup and validation loop repeatable.
From the repository root:
python -m pip install -e .Create a starter project:
mcpforge init ./my-serverThat command generates:
server.pywith a minimal stdio MCP serverREADME.mdwith local run instructions.gitignorefor basic Python artifacts
Run the generated server locally:
cd my-server
python server.pyIn another terminal, validate the server:
mcpforge check .Successful output looks like:
PASS: stdio server handshake succeeded.
Server: starter-mcp-server 0.1.0
Verified methods: initialize, tools/list, resources/list, prompts/list
The current implementation is intentionally small:
- one scaffold command
- one local stdio smoke-test command
- one minimal project template
That is enough to make the repo usable while keeping the first implementation easy to understand and extend.
Likely next improvements include:
- richer starter templates
- configurable server names and metadata
- deeper MCP protocol validation
- CI-friendly check output
- packaging and deployment helpers
MCPForge/
├── .gitignore
├── README.md
├── pyproject.toml
└── mcpforge/
├── __init__.py
├── cli.py
└── templates.py