Maestro is an AI build verification tool that runs a 14-point Guardian pipeline on your code. Catches bugs Cursor, Claude Code, and Copilot miss. Auto-fixes what it can. Guarantees output works before delivery.
$ pip install maestro-guard && maestro-guard check index.html
AI coding tools generate great code — but they also generate silent, subtle bugs. Maestro finds them before they reach production.
Catches IDs, classes, and selectors referenced in JS but missing from HTML. The #1 hidden bug in AI-generated UIs.
Detects leftover console.errors, debug logs, and dev-only code that LLMs leave behind. Blocks them from shipping.
Flags inline event handlers, innerHTML usage, and missing CSP headers — the most common AI-generated security gaps.
LLM-powered verifier reads your original spec and checks if the output actually implements what was asked — not just "does it compile?"
15+ bug patterns auto-detected and fixed: MOCK_DATA left in production, template literals broken, e.target_N patterns, and more.
Each build is reviewed by 5 AI agents: Security, UX, Code Quality, SEO, and Business Viability — with heuristic fallback when LLMs are unavailable.
Three commands. One pipeline. Zero guesswork.
Define what you want to build in a simple markdown spec. Maestro uses it as ground truth for verification.
Use Cursor, Claude Code, Copilot, or Codex — whatever generates the best code for you. Maestro is tool-agnostic.
Run maestro-guard check — 14 checks, auto-fixes, and spec fulfillment gate. Pass or don't ship.
Every check must pass at 100/100 or delivery is blocked. Here's what runs on every file:
Maestro Guard weighs nothing. Pure Python stdlib. No npm, no Docker, no cloud service. Drop it in any CI/CD pipeline.
Maestro catches what your AI coding tool misses. One command. Every build. Guaranteed.
📦 Install Maestro →pip install maestro-guard · Open Source · MIT