paiml -agent-toolkit

PMAT is a multi-purpose agent toolkit that includes an MCP server to expose its AI-assisted code analysis capabilities to other tools and systems. When you run the MCP server, it provides a collection of endpoints and prompts that let clients request AI-generated context, code quality insights, and governance checks across your codebase. The toolkit can perform context generation, technical debt grading, mutation testing, repository health scoring, and a suite of workflow prompts, all designed to support automation pipelines and AI-assisted development workflows. Use the MCP server to enable Claude Code, Cline, and other agents to query your repository’s state, fetch rich code-context outputs, and enforce quality gates during CI/CD or in agent-driven automation. Start the MCP server with the built-in CLI (pmat mcp) and then connect your MCP clients to it to issue requests like context generation, TDG scoring, repo scoring, and more.

Key capabilities exposed by the MCP server include: generating AI-ready context for codebases, analyzing code quality and technical debt across projects, performing mutation testing to validate tests, scoring repository health on a quantitative scale, and applying a broad set of governance and compliance checks. The server also exposes workflow prompts for enforcing quality gates and supports cross-tool integration with agents and pipelines.

To use the server’s tools, connect via your MCP client to request specific tasks such as generating context, running tdg analyses, or performing repository health assessments. You can tailor prompts and outputs to your automation needs, retrieve structured results, and pipe them into CI workflows or reporting dashboards.

Prerequisites:

• A working Rust toolchain (Rustup, cargo) or a prebuilt release binary of PMAT.
• Internet access to install dependencies and fetch crates.

Option A: Install from crates.io (recommended for development and updates)

1. Install Rust and Cargo from https://www.rust-lang.org/tools/install
2. Install PMAT binary: cargo install pmat
3. Verify installation: pmat --version

Option B: Install from source

1. Clone the repository: git clone https://github.com/paiml/paiml-mcp-agent-toolkit
2. Build and install: cd paiml-mcp-agent-toolkit cargo install --path .
3. Verify installation: pmat --version

Start the MCP server:

• Run the server (using the built-in MCP command): pmat mcp
• By default, the server will start and listen for MCP clients on the configured port. If you need to customize the host/port, refer to your environment or the CLI options for pmat mcp (e.g., --host, --port) and adjust accordingly.

Tips and notes:

• This MCP server is Rust-based and is started via the PMAT CLI (pmat mcp). Ensure your environment has network access to allow clients to connect.
• If you need to customize behavior or endpoints, consult the PMAT configuration and any environment variables documented in the project or via --help on the mcp command.
• For CI/CD integrations, consider exposing the MCP server behind a secure gateway or using authentication mechanisms supported by your deployment environment.
• Common issues include version mismatches between the PMAT client and server; ensure both sides are compatible by using the same PMAT release.
• If you deploy in containers or orchestration platforms, you can wrap the PMAT startup in your container entrypoint and expose the necessary port for MCP traffic.