| name | feinblick |
| description | Audit a repo for dead code, cycles, complexity, and broken Claude skills, then read the findings. Use when verifying code or skill health in CI or an agent loop. |
feinblick — codebase intelligence
feinblick is a command on your PATH. It unifies Python static analysis
(CytoScnPy, Tach) and Claude-skill validation into one finding model, an audit
gate, and a health score. Call it as a bare command — never a file path or cd.
In CI or an agent loop, gate on findings introduced since the base branch and
read the machine output:
feinblick audit --changed-since origin/main --format json
Each entry in findings[] carries actions[] with an auto_fixable flag and,
when available, an engine_fix_cmd — act on those first.
For a full, un-gated report over code, skills, or both:
feinblick check code
For just the 0-100 health score and its hotspots:
feinblick health
audit exits non-zero only when the gate fails, so it is safe to wire directly
into a pipeline step.