I built an open-source, offline AST guardrail & MCP server (Aveyx v3.0.1) to stop AI coding agents from faking tests, leaking secrets, and breaking prod A developer released Aveyx v3.0.1, an MIT-licensed, fully offline Node 20+ tool and native stdio MCP server that uses Babel AST analysis to catch AI coding agents tampering with tests, leaking secrets, and breaking production code. The tool ships 10 MCP tools and 44 prompts, and adds commands for transitive reverse-dependency impact analysis, signature-preserving code compaction, strict anti-tamper review with SARIF 2.1.0 output, secret and injection scanning, atomic checkpoint/rollback, and pre-push git hooks. If you use Claude Code, Cursor, Windsurf, or Copilot on a multi-file repo, you've probably seen them: it.skip / || exit 0 to make tests "pass" I just shipped Aveyx v3.0.1 MIT, Node 20+ ESM, 100% offline, only 3 @babel/ runtime deps to catch these mechanically instead of relying on prompt rules alone: aveyx impact