Package Version Check

You can use this MCP server to fetch the latest, up-to-date versions of packages and development tools across many ecosystems. It helps you replace outdated pins in AI-generated code and CI configurations, ensuring you always reference the current versions available from the registries and tooling ecosystems you rely on.

How to use

You access the MCP server through a client that can call its tools. The primary capabilities let you fetch the latest package versions across ecosystems like NPM, PyPI, NuGet, Maven/Gradle, Go, PHP, Ruby, Rust, Swift, Dart, Docker images, Helm, Terraform providers and modules, as well as GitHub Actions metadata and a wide set of development tools managed by mise-en-place. Use the MCP to verify and substitute dependency versions in your prompts, configuration files, or automation scripts.

Key usage patterns include: finding latest package versions before committing changes, pinning tool versions in wrapper or config files (Gradle, Maven, Terraform, kubectl, etc.), and retrieving GitHub Actions versions and inputs/outputs when you rely on Actions hosted on GitHub.com. When your AI-generated code or CI workflows pin versions, run the MCP tools to refresh those pins with the current versions. Then, update your files to reflect the latest versions.

How to install

You have three options to run the MCP server and connect your agent.

Option 1: Use the Hosted Service (Easiest) — HTTP mode

https://package-version-check-mcp.onrender.com/mcp

Option 2: Run with uvx (for local use)

Use the uvx runner to execute the MCP server locally. This combines ease of use with automatic installation of the latest version from PyPI.

uvx package-version-check-mcp --mode=stdio

Option 3: Run with Docker (for local use)

Run the pre-built Docker image to host the MCP server locally.

docker run --pull=always --rm -i ghcr.io/mshekow/package-version-check-mcp:latest --mode=stdio

Caching configuration

To improve performance and reduce requests to registries, you can enable in-memory caching with TTL. Configure these environment settings in your deployment.

PACKAGE_VERSION_CACHE_ENABLED=true
PACKAGE_VERSION_CACHE_TTL_SECONDS=3600
PACKAGE_VERSION_CACHE_MAX_SIZE_MB=64

Configuring your agent

After adding the MCP server, enable its tools in your agent configuration. Then guide your agent to use the MCP when generating or updating code by including prompts such as:

• Use MCP to get latest versions
• Check the latest package versions using the MCP tools
• Use get_latest_package_versions to find the current version

Notes and considerations

The hosted service may have rate limits from the underlying registries. Local runs via uvx or Docker provide a private, self-hosted option. If you rely on GitHub APIs, you can reduce rate limits by providing a GitHub Personal Access Token via the GITHUB_PAT environment variable.

Development references

For development or testing, you can run the MCP server locally in a Python environment. Follow the project’s setup steps to prepare a virtual environment and install dependencies, then start the server with the provided entry point.

Tools and endpoints overview

The MCP exposes several tooling endpoints to fetch the latest versions and metadata. Use them to integrate up-to-date values into your workflows and AI-generated content.

Available tools

get_latest_package_versions

Fetches the latest versions for packages across ecosystems such as npm, pypi, nuget, maven_gradle, go, php, rubygems, rust, swift, dart, docker, helm, terraform_provider, and terraform_module.

get_github_action_versions_and_args

Fetches the latest versions and metadata for GitHub Actions hosted on github.com, including option to include the action README.

get_supported_tools

Returns a list of all tool names supported by the mise-en-place tool registry.

get_latest_tool_versions

Fetches the latest stable versions of development and DevOps tools supported by mise-en-place, outside of language ecosystems.