@codecheck/mcp-server

MCP server for CodeCheck — lets Claude Code and any MCP-compatible AI assistant run AI-generated tests directly from within their workflow, without you asking.

When this server is connected, Claude can call codecheck_run after editing your files, see which functions passed and failed, and fix issues before you even commit.

Install in Claude Code

Add to your Claude Code settings (~/.claude/settings.json for global, or .claude/settings.json for a single project):

{
  "mcpServers": {
    "codecheck": {
      "command": "npx",
      "args": ["-y", "@codecheck/mcp-server"],
      "env": {
        "ANTHROPIC_API_KEY": "sk-ant-..."
      }
    }
  }
}

That's it. Restart Claude Code and it will have access to three new tools.

Tools

codecheck_run

Run AI-generated tests on source files. Extracts every function, sends them to your configured LLM, generates test cases, runs them, and returns a pass/fail report.

codecheck_run({ files: ["src/utils.ts", "src/payment.ts"] })

→ CodeCheck Results
  ──────────────────────────────────────────────────
  4 passed · 1 failed · 80% pass rate

    src/payment.ts
      ✓ processPayment — unit · happy-path  (14ms)
      ✓ processPayment — unit · type-error  (11ms)
      ✓ processPayment — smoke · happy-path  (9ms)
      ✗ processPayment — unit · null-check  (8ms)
        → Crashed on null or undefined input — the function needs a null check
          TypeError: Cannot read properties of null

Parameters:

• files (required) — array of source file paths (absolute or relative to project root)
• projectDir — project root, defaults to the server's working directory
• testTypes — override test types: unit, smoke, functional, sanity, integration, api, e2e, snapshot, regression

codecheck_results

Get the results from the most recent run without re-running tests.

codecheck_status

Check whether CodeCheck is configured in the project, which AI provider is in use, and whether the API key is set.

Providers

The MCP server uses the same provider configured in your project's codecheck config (package.json or codecheck.config.json).

| Provider | Env var to set in MCP config | |---|---| | Anthropic (default) | ANTHROPIC_API_KEY | | OpenAI | OPENAI_API_KEY | | Gemini | GEMINI_API_KEY | | Ollama | OLLAMA_BASE_URL (optional, defaults to http://localhost:11434/v1) |

Project setup

The project being tested needs a codecheck config. The fastest way:

npm install -D @codecheck/trigger-oncommit @codecheck/scope-unit @codecheck/output-terminal
npx codecheck-init

Or add manually to package.json:

{
  "codecheck": {
    "provider": "anthropic",
    "model": "claude-sonnet-4-6",
    "language": "typescript",
    "framework": "vitest",
    "testTypes": ["unit", "smoke"],
    "threshold": 0.8,
    "failOnError": false
  }
}

Links

• CodeCheck ecosystem
• All @codecheck packages
• MCP protocol

License

MIT © Medhavee Upadhyaya