500+ AI agents exist, but no way to know which ones actually work. Benchmarks evaluate LLMs, not agents. Two agents on the same GPT-4o can have wildly different reliability. Legit evaluates agents, not models. 36 tasks, 3 AI judges (Claude + GPT-4o + Gemini), one trust score. Three commands. Zero cost. Five minutes. Open source, Apache 2.0.
No reviews yetBe the first to leave a review for Legit
Maker
📌
Hey everyone! We built Legit because we kept picking AI agents that
looked great in demos but fell apart in production. Existing benchmarks
tell you how good the LLM is, but nothing about the agent built on top.
Legit scores the full agent stack -- prompts, tool use, error handling,
orchestration. Three independent AI judges evaluate your agent and take
the median score to prevent bias.
Would love your feedback on the scoring methodology.
▶︎GitHub: github.com/getlegitdev/legit