tools-executor 0.1.0

Run a FastMCP proxy that lists and invokes every MCP tool from the CLI or an HTTP server.

tools-executor

tools-executor is a lightweight FastMCP driver that discovers every MCP tool exposed by your configured backends, prints Python-like signatures, and lets you invoke them either from the command line or through a minimal FastMCP server.

Why use it?

• 📋 Discoverability: Generate readable signatures for every tool so you know which arguments to pass.
• ⚡ Fast iteration: Jump between running the HTTP/stdio server and issuing one-off tool calls without reconfiguring clients.
• 🧰 Batteries included: Ships with a uv-first workflow for building, testing, and publishing to PyPI.

Installation

pip install tools-executor

or, if you prefer the uv toolchain:

uv tool install tools-executor

Configuration

Point the CLI at an MCP client definition. Copy mcp.json.example into your workspace, adjust the upstream server definitions, and pass the path with -c/--config.

cp mcp.json.example ~/.config/tools-executor/mcp.json

Usage

List available tools and inspect their signatures:

tools-executor list -c ~/.config/tools-executor/mcp.json

Call a tool directly:

tools-executor call -c ~/.config/tools-executor/mcp.json weather --arg city="Lisbon"

Run the FastMCP server (HTTP by default):

tools-executor serve -c ~/.config/tools-executor/mcp.json --transport http --host 0.0.0.0 --port 23456

Every command shares the --config option so you can point at different MCP client definitions per invocation.

Local development

uv venv --seed
uv sync
uv run tools-executor list -c mcp.json.example

The repository still ships main.py so you can run ./main.py list ... directly with uv run if you prefer scripting locally.

Releasing with uv

1. Bump version inside pyproject.toml.
2. Build and verify the artifacts:

   uv build
   uv run python -m tools_executor.cli --help
   

3. Publish to PyPI with an API token (create one under your PyPI account and store it securely):

   export PYPI_API_TOKEN="pypi-xxxxxxxxxxxxxxxxxxxxxxxx"
   uv publish --token "$PYPI_API_TOKEN"
   

4. Tag the release in git and push (git tag v0.1.1 && git push --tags).

GitHub Action publish workflow

The repository includes .github/workflows/publish.yml, which builds the wheel and sdist via uv build and calls uv publish --token $PYPI_API_TOKEN. Add a PYPI_API_TOKEN secret (scoped to “Publish to PyPI”) in your repository settings and trigger the workflow from the Actions tab or by creating a GitHub Release.