> ## Documentation Index > Fetch the complete documentation index at: https://docs.scanoss.com/llms.txt > Use this file to discover all available pages before exploring further. # Earnie for AI Architects > Who governs the agents: grounded, deterministic answers over MCP before a coding agent's work lands. ## "Who Governs the Agents?" Before a coding agent hands code back, it can ask Earnie what's allowed and check its own work over [MCP](/en/latest/earnie/using-earnie/earnie-cli#connecting-a-coding-agent-mcp), at the moment of the decision rather than after the fact. The answer is grounded in what the code actually matches, not model guesswork, and it's the same deterministic verdict a human reviewer would get from the same input. ## MCP Integration [`earnie mcp setup`](/en/latest/earnie/using-earnie/earnie-cli#setup) connects Claude Code, Cursor, VS Code, or Codex to Earnie with one API key. Once connected, an agent's end-of-turn review is a [Self-Check](/en/latest/earnie/getting-started/first-scan#self-checks-the-dry-run-before-the-gate): it costs nothing, creates no findings, and moves no posture, so an agent can check risky code in context, before it lands, as often as it needs to. **AI model detection and the AI-BOM are not yet available.** Detecting embedded AI SDKs, model files, and AI API keys, and exporting an AI-BOM mapped to frameworks like the EU AI Act or NIST AI RMF, is on the roadmap but isn't part of the shipped product today. This page will be updated once that capability ships. ## What's Next Read [Connecting a Coding Agent (MCP)](/en/latest/earnie/using-earnie/earnie-cli#connecting-a-coding-agent-mcp) for the full setup, or [Earnie for AI Agents](/en/latest/earnie/by-role/ai-agents) for what an agent can assume about the answers it gets back.