Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Log all errors with full context, detect patterns, and suggest approach mutations to avoid repeated failures.
.claude/skills/a5c-ai-error-logging/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 160% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 69% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -57% | 0% |
Log all errors with full context, detect patterns, and suggest approach mutations to avoid repeated failures.
Error Analyst - pwf-error-analyst
projectPath - Root path for planning filesphaseName - Phase where error occurrederrors - Array of error objectserrorHistory - Previous errors for pattern detectionplanning-orchestrator.js - Error logging during phasesplanning-execution.js - Immediate error logging, pattern analysisplanning-verification.js - Error resolution checking| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 37,862 | 6,159 | -84% | 1 | 1 | 0% | 6,348 | 573 | -91% | 0 | 0 | — |
case-02 | fail→pass | 6,436 | 14,501 | +125% | 1 | 1 | 0% | 1,090 | 2,839 | +160% | 0 | 0 | — |
case-03 | fail→pass | 9,265 | 12,597 | +36% | 1 | 1 | 0% | 1,432 | 2,416 | +69% | 0 | 0 | — |
case-04 | fail→fail | 3,920 | 4,895 | +25% | 1 | 1 | 0% | 175 | 463 | +165% | 0 | 0 | — |
case-05 | fail→fail | 4,714 | 4,458 | -5% | 1 | 1 | 0% | 957 | 1,140 | +19% | 0 | 0 | — |
case-06 | fail→fail | 4,188 | 4,390 | +5% | 1 | 1 | 0% | 848 | 1,032 | +22% | 0 | 0 | — |
case-07 | fail→fail | 8,192 | 7,133 | -13% | 1 | 1 | 0% | 1,319 | 1,364 | +3% | 0 | 0 | — |
case-08 | fail→pass | 4,563 | 2,446 | -46% | 1 | 1 | 0% | 785 | 720 | -8% | 0 | 0 | — |
case-09 | fail→pass | 8,474 | 2,044 | -76% | 1 | 1 | 0% | 1,534 | 506 | -67% | 0 | 0 | — |
case-10 | fail→pass | 10,417 | 2,944 | -72% | 1 | 1 | 0% | 1,811 | 779 | -57% | 0 | 0 | — |
case-11 | fail→pass | 3,176 | 2,869 | -10% | 1 | 1 | 0% | 450 | 686 | +52% | 0 | 0 | — |
case-12 | fail→pass | 11,484 | 1,914 | -83% | 1 | 1 | 0% | 1,898 | 596 | -69% | 0 | 0 | — |
case-13 | fail→pass | 7,463 | 953 | -87% | 1 | 1 | 0% | 1,228 | 364 | -70% | 0 | 0 | — |
case-14 | fail→pass | 12,964 | 1,428 | -89% | 1 | 1 | 0% | 990 | 474 | -52% | 0 | 0 | — |
case-15 | fail→pass | 7,817 | 1,159 | -85% | 1 | 1 | 0% | 1,385 | 381 | -72% | 0 | 0 | — |
case-16 | fail→fail | 8,340 | 2,350 | -72% | 1 | 1 | 0% | 1,210 | 611 | -50% | 0 | 0 | — |
case-17 | fail→fail | 4,097 | 2,698 | -34% | 1 | 1 | 0% | 658 | 661 | +0% | 0 | 0 | — |
case-18 | fail→pass | 5,719 | 1,928 | -66% | 1 | 1 | 0% | 914 | 524 | -43% | 0 | 0 | — |
case-19 | fail→fail | 5,133 | 5,993 | +17% | 1 | 1 | 0% | 726 | 1,169 | +61% | 0 | 0 | — |
case-20 | fail→fail | 10,283 | 3,836 | -63% | 1 | 1 | 0% | 1,542 | 906 | -41% | 0 | 0 | — |
case-21 | fail→pass | 2,452 | 1,533 | -37% | 1 | 1 | 0% | 347 | 410 | +18% | 0 | 0 | — |
case-22 | pass→pass | 14,135 | 1,544 | -89% | 1 | 1 | 0% | 2,361 | 474 | -80% | 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 19 counted toward the lift figure. The other 3 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 +55 percentage points is the difference between those two pass rates over the 19 comparable cases.
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.