Evals & guardrails for AI agents and the MCP tools they call: lint + auto-fix, fuzz, SHIP/FIX-FIRST/BLOCK verdict, live guard. Runs in CI, Claude Code, Codex, Cursor.