paiml-mcp-agent-toolkit

概述

实用多语言代理工具包 (PMAT) 通过其 MCP 服务器实现提供了全面的代码分析功能。它为 AI 代理提供 19 种工具，用于分析代码质量、生成技术债务评分、执行变异测试和创建 AI 优化上下文。使用 Rust 构建，具有广泛的测试覆盖率 (99.66%) 和积极的维护，PMAT 与 Claude Code 和其他 AI 编码助手集成。该工具包遵循丰田生产方式质量原则，包含 Git 历史 RAG、语义搜索和合规治理等功能，支持 20 多种编程语言。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• context

  Generate AI-ready context for code analysis

  试试:Generate LLM-optimized context for this codebase试试:Create context including test files
• analyze

  Analyze code quality metrics including complexity and technical debt

  试试:Analyze technical debt in the project试试:Check code complexity issues
• repo-score

  Score repository health across 11 categories (0-289 scale)

  试试:Score the repository health试试:Get detailed repository metrics
• mutate

  Run mutation testing to validate test suite effectiveness

  试试:Run mutation testing on the src directory试试:Check test coverage with mutation testing
• query

  Search code with semantic analysis and git history

  试试:Search for "error handling patterns"试试:Find functions related to cache invalidation
• comply

  Run compliance governance checks across code quality and best practices

  试试:Run all compliance checks试试:Check for technical debt regression
• kaizen

  Autonomous continuous improvement - scan, auto-fix, and commit changes

  试试:Run kaizen analysis with dry-run试试:Apply auto-fixes and commit improvements
• extract

  Extract function boundaries with metadata

  试试:Extract functions from src/lib.rs试试:List all functions with visibility info
• infra-score

  Score CI/CD infrastructure quality (0-100 scale)

  试试:Evaluate CI/CD pipeline quality试试:Get infrastructure health report
• hooks

  Manage Git hooks for quality enforcement

  试试:Install pre-commit hooks试试:Check hook installation status
• prompt

  Run pre-configured AI prompts for TDD and quality enforcement

  试试:List available prompts试试:Run code coverage enforcement prompt
• tdg-baseline

  Create technical debt baseline and check for regressions

  试试:Create TDG baseline试试:Check for quality score degradation

可对比工具

安装

cargo install pmat

pmat mcp

{
  "mcpServers": {
    "pmat": {
      "command": "pmat",
      "args": ["mcp"]
    }
  }
}

FAQ

PMAT 支持哪些编程语言？

PMAT 支持 20 多种编程语言，包括 Rust、TypeScript、Python、Go、Java、C/C++、Lua、Lean 以及各种 MLOps 模型格式。

PMAT 如何确保自身代码库的质量？

PMAT 遵循丰田生产方式原则，拥有 21,200+ 个通过的测试、99.66% 的覆盖率、>80% 的变异评分，并通过 CI 强制执行明确的质量承诺。

paiml-mcp-agent-toolkit 对比