Skip to main content
Glama
README.md
# GauntletCI MCP Server

Model Context Protocol (MCP) server that exposes GauntletCI behavioral change risk analysis
as tools for AI coding assistants (Claude, Copilot, Cursor, etc.).

## Tools

| Tool | Description |
|------|-------------|
| `analyze_commit` | Run GauntletCI on HEAD, return findings as readable text |
| `get_findings_json` | Run GauntletCI, return raw JSON for programmatic use |
| `get_sarif` | Run GauntletCI, return SARIF 2.1.0 report |

## Requirements

- Node.js 20+
- GauntletCI CLI: `dotnet tool install -g GauntletCI`

## Installation

```bash
npm install
npm run build
```

## Usage with Claude Desktop

Add to `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "gauntletci": {
      "command": "node",
      "args": ["/path/to/GauntletCI-MCP/dist/index.js"]
    }
  }
}
```

## Usage with VS Code (Copilot)

Add to `.vscode/mcp.json`:

```json
{
  "servers": {
    "gauntletci": {
      "type": "stdio",
      "command": "node",
      "args": ["${workspaceFolder}/path/to/GauntletCI-MCP/dist/index.js"]
    }
  }
}
```

## Example prompts

Once connected, ask your AI assistant:

- "Run GauntletCI on this repo at /home/user/myapp and summarize the findings."
- "Are there any block-level risks in the current commit?"
- "Get the SARIF report for this commit."

## Links

- [MCP specification](https://modelcontextprotocol.io)
- [GauntletCI documentation](https://gauntletci.com/docs)
- [GauntletCI engine](https://github.com/EricCogen/GauntletCI)

TDQS

B3.1/5.0

Scored across 3 tools

Disambiguation4/5

The three tools all trigger the same GauntletCI analysis but return different outputs (structured text, JSON, SARIF). The descriptions clearly differentiate the output formats, making it easy for an agent to choose based on the desired result.

Naming Consistency2/5

Naming is inconsistent: 'analyze_commit' uses a verb_noun pattern while 'get_findings_json' and 'get_sarif' use 'get_'. This mixed convention could cause confusion about the action of each tool.

Tool Count4/5

Three tools is a lean but appropriate set for a focused CI analysis server. Each output format covers a distinct use case without unnecessary bloat.

Completeness2/5

The tool surface is incomplete: it only analyzes the current HEAD commit and lacks options for specifying a different commit, configuring analysis parameters, or retrieving previous analyses. This leaves significant gaps for users needing more control.

Maintenance

ActivityInactive
ResponsivenessNo issues