mcp -pacman

Pacman is a Model Context Protocol server that exposes a flexible set of tools to search and inspect package indices and container registries. It aggregates providers for PyPI, npm, crates.io, Docker Hub, and Terraform Registry, enabling you to query packages, retrieve metadata, and fetch detailed information about specific items. The main tools you can invoke are: search_package for querying package indices, package_info for package details, search_docker_image and docker_image_info for Docker Hub, and terraform_module_latest_version for Terraform modules. Use the appropriate prompts to issue structured requests to the server and receive concise results suitable for feeding into an LLM or downstream tooling. The server is designed to be run locally or in a container and can be integrated into Claude.app, VS Code, or other orchestration environments via the provided configuration examples.

Prerequisites:

• Python (for uvx-based usage) or uvx tooling installed
• Access to the internet to fetch package index data
• Docker (optional, for container usage)

Installation options:

1. Using uvx (recommended):

• Ensure uv and uvx are installed following the project guidance, then run:

uvx mcp-server-pacman

2. Using PIP:

• Install the server package via pip:

pip install mcp-server-pacman

• Run the server as a module:

python -m mcp_server_pacman

3. Using Docker:

• Pull and run the image:

docker pull oborchers/mcp-server-pacman:latest
docker run -i --rm oborchers/mcp-server-pacman

If you prefer explicit command-line control, you can point uvx or uv at the installed package as demonstrated in the Usage section of the README and configure your environment accordingly.

Tips and caveats:

• You can customize the user-agent by appending --user-agent to the server arguments in your MCP configuration.
• When using Docker, ensure the container has network access to the target package indices.
• The server supports a range of prompts for different providers (pypi, npm, crates, docker, terraform) and corresponding info queries. If a provider lacks data, handle empty results gracefully in downstream tooling.
• For debugging, you can use the MCP inspector tools as described in the README to validate requests and responses.
• If you need to run tests or lint locally, follow the uv-based development commands referenced in the repository documentation.