Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Audits repositories for Claude Code readiness and suggests improvements. Use when asked to check CLAUDE.md quality, review settings, audit project organization, or optimize for agentic work.
.claude/skills/bilal140202-optimizing-claude-code/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -16% | 0% |
Default: Audit-only. Present findings and await approval before modifying files.
Execute from repo root:
bashpython scripts/audit_repo.py --root . --include-user-scope
Analyzes:
Structure output as:
Executive Summary (5 bullets max)
Repository Profile
Memory Files Assessment
Settings & Configuration
Issues by Priority
Recommendations
When asked to implement:
When file count exceeds 3000, include scaling recommendations:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 16,794 | 5,984 | -64% | 1 | 1 | 0% | 2,792 | 673 | -76% | 0 | 0 | — |
case-02 | fail→fail | 21,131 | 5,035 | -76% | 1 | 1 | 0% | 3,238 | 734 | -77% | 0 | 0 | — |
case-03 | fail→fail | 24,292 | 6,370 | -74% | 1 | 1 | 0% | 3,968 | 737 | -81% | 0 | 0 | — |
case-04 | pass→fail | 14,869 | 5,853 | -61% | 1 | 1 | 0% | 2,963 | 769 | -74% | 0 | 0 | — |
case-05 | pass→pass | 8,596 | 10,410 | +21% | 1 | 1 | 0% | 1,499 | 2,206 | +47% | 0 | 0 | — |
case-06 | pass→pass | 9,459 | 10,322 | +9% | 1 | 1 | 0% | 1,829 | 2,372 | +30% | 0 | 0 | — |
case-07 | fail→pass | 14,393 | 6,137 | -57% | 1 | 1 | 0% | 2,349 | 802 | -66% | 0 | 0 | — |
case-08 | pass→pass | 7,733 | 2,819 | -64% | 1 | 1 | 0% | 1,153 | 773 | -33% | 0 | 0 | — |
case-09 | fail→fail | 12,039 | 3,082 | -74% | 1 | 1 | 0% | 1,893 | 926 | -51% | 0 | 0 | — |
case-10 | pass→pass | 8,273 | 2,434 | -71% | 1 | 1 | 0% | 1,211 | 694 | -43% | 0 | 0 | — |
case-11 | fail→pass | 13,034 | 10,758 | -17% | 1 | 1 | 0% | 1,936 | 1,975 | +2% | 0 | 0 | — |
case-12 | fail→pass | 17,398 | 3,132 | -82% | 1 | 1 | 0% | 2,661 | 909 | -66% | 0 | 0 | — |
case-13 | pass→pass | 12,938 | 6,943 | -46% | 1 | 1 | 0% | 1,891 | 1,575 | -17% | 0 | 0 | — |
case-14 | fail→pass | 7,480 | 3,227 | -57% | 1 | 1 | 0% | 1,183 | 922 | -22% | 0 | 0 | — |
case-15 | pass→pass | 13,366 | 5,036 | -62% | 1 | 1 | 0% | 2,090 | 1,262 | -40% | 0 | 0 | — |
case-16 | fail→pass | 14,779 | 9,514 | -36% | 1 | 1 | 0% | 2,253 | 1,897 | -16% | 0 | 0 | — |
case-17 | fail→pass | 18,275 | 7,651 | -58% | 1 | 1 | 0% | 2,578 | 1,624 | -37% | 0 | 0 | — |
case-18 | fail→pass | 16,479 | 11,132 | -32% | 1 | 1 | 0% | 2,368 | 2,014 | -15% | 0 | 0 | — |
case-19 | fail→pass | 9,958 | 2,082 | -79% | 1 | 1 | 0% | 1,542 | 787 | -49% | 0 | 0 | — |
case-20 | pass→fail | 12,763 | 4,965 | -61% | 1 | 1 | 0% | 1,885 | 986 | -48% | 0 | 0 | — |
case-21 | pass→pass | 15,032 | 6,358 | -58% | 1 | 1 | 0% | 2,235 | 1,484 | -34% | 0 | 0 | — |
case-22 | pass→pass | 9,679 | 2,058 | -79% | 1 | 1 | 0% | 1,604 | 738 | -54% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 18 counted toward the lift figure. The other 4 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +27 percentage points is the difference between those two pass rates over the 18 comparable cases. 5 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.