Print the per-critic agreed/availability scorecard and drop advice. Use for CR critic scores or /cr-scores.
Install
npx skillscat add yotamleo-org/himmel/cr-scores Install via the SkillsCat registry.
SKILL.md
cr-scores
When the user asks for the critic scorecard, run:
bash scripts/cr/cr-scores.sh [--window N]--window N narrows to the last N PRs (default 20). Summarise the all-time +
windowed per-model table (total findings, agreed % — fixed verdicts count as
agreed, disproved %, conflict count, unaddressed count, availability %) and
surface any "consider dropping" lines verbatim. If the ledger is empty, report
"no critic scores recorded yet" and stop.
See .claude/commands/cr-scores.md.