It analyzes a GitHub repo (URL or local path) and outputs: - an architecture/dependency map (+ hotspots) - runnable quickstart guidance - risks (license/CI/config signals) - actionable issues (incl. “good first”)
Killer feature: with --try-run, it attempts real install/test/build/start steps in a sandbox and attaches evidence/timeouts to the report bundle.
Repo + demo GIF: I’ll add in the first comment. Feedback welcome on: try-run reliability, report format, and what heuristics you’d want next.