MCP Integration

The Faultline MCP server lets AI agents (Claude, Cursor, Copilot) query your Faultline Enterprise data — risk scores, findings, suppression reports, and audit events — without leaving their workflow.

MCP endpoint

POST https://mcp.gofaultline.dev/mcp

Streamable HTTP transport using the standard MCP initialize handshake. No server-level API key required.

Connect to Claude.ai

1. Open Claude.ai → Settings → Integrations
2. Click Add MCP Server
3. Enter: https://mcp.gofaultline.dev/mcp
4. No server key — authentication is per tool call

Connect to Cursor

Add to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "faultline": {
      "url": "https://mcp.gofaultline.dev/mcp"
    }
  }
}

Authentication

Each tool call requires a Faultline API token passed as the api_token argument. Create a token at app.gofaultline.dev → Settings → API Tokens.

Two tools require no authentication: faultline_explain_finding and faultline_explain_score.

Available tools

18 tools across two categories:

• Enterprise tools (14) — org summary, executive dashboard, repo list, packages, findings, trends, regression attribution, suppression report, ownership report, owner scorecard, policy packs, audit events, repo suppressions, PR advisory
• Utility tools (2) — explain finding code, explain risk score

Example prompts

Using Faultline, get the risk summary for org [org_id]
with API token [token], then tell me which packages are highest risk.

What does FL-OWN-003 mean and how do I fix it?

Show me suppression governance report for org [org_id]
— how many are expiring this month?

Agent Skills

The Faultline skill is available for Agent Skills-compatible clients at:

https://skills.gofaultline.dev/faultline/SKILL.md

Discovery index: https://skills.gofaultline.dev/llms.txt