root@studio:~$ cd
Tracing route… ///
Tracing route… ///
Paste your agent config, AGENTS.md, skills files, CLAUDE.md, and get a verdict in seconds: capability versus process, missing practices, and what to add first. Runs entirely in your browser.
Local audit runs entirely in your browser. The AI audit sends your config to Groq.
Every line of your config is classified as capability (“how things work”) or process (“how we think”). Capability lines make the agent faster; process lines keep it honest. The audit also checks for the four practices that separate a judgment library from an API map: a load-bearing rule, a why attached to rules, a postmortem-to-diff rule, and a prune schedule.
Read the argument