Analyze, score, benchmark, and store the files that define how your AI agents work. Know exactly what’s strong, what’s weak, and what to fix.
We ingest the docs first. We ask the follow-up questions after the files are in.
CLAUDE.md, AGENTS.md, .cursorrules, prompt docs, guardrails, routing files, tool docs
No account required for evaluation. Sign in only if you want to save the system as a collection.
Works with every major AI coding platform
Three steps to measurably better agent instructions.
Paste or drop any agent instruction file. CLAUDE.md, .cursorrules, AGENTS.md, or any custom format.
20-point rubric plus AI semantic analysis across Structure, Reasoning, Control, and Engagement.
Actionable recommendations with benchmarking proof. See exactly what to fix and why.
From instant scoring to production storage. One platform for the full lifecycle.
Deterministic scoring across Structure, Reasoning, Control, and Engagement axes with star tier rankings.
Test your agent file against Claude, GPT-4o, Gemini, and more to find the optimal model fit.
Progressive enrichment from Bronze to Platinum with response testing proof.
Store agent files externally with version history, drift protection, and lockdown mode.
Serve your agent files to any LLM client via Model Context Protocol or traditional REST API.
Check your file against 17 codified recommendations from Anthropic, OpenAI, Google, and more.
Rigorous, deterministic quality measurement you can trust.
Start free. Upgrade when you need production features.
Engineering notes, launch updates, and how we think about prompt quality.
What shipped recently, what improved, and what changed in the platform.
Find out how your agent file scores in 30 seconds.