Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Post-mortem investigation for failed GSD workflows — analyzes git history, artifacts, and state to diagnose what went wrong
.claude/skills/davepoon-gsd-forensics/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-14 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-05 | ✓→✗ | ▼ Worse | -60% | 0% |
| case-11 | ✓→✗ | ▼ Worse | -69% | 0% |
| case-12 | ✓→✗ | ▼ Worse | 63% | 0% |
<objective> Investigate what went wrong during a GSD workflow execution. Analyzes git history, .planning/ artifacts, and file system state to detect anomalies and generate a structured diagnostic report.
Purpose: Diagnose failed or stuck workflows so the user can understand root cause and take corrective action. Output: Forensic report saved to .planning/forensics/, presented inline, with optional issue creation. </objective>
<execution_context> @${CLAUDE_PLUGIN_ROOT}/workflows/forensics.md </execution_context>
<context> Data sources:
git log (recent commits, patterns, time gaps)git status / git diff (uncommitted work, conflicts).planning/STATE.md (current position, session history).planning/ROADMAP.md (phase scope and progress).planning/phases/*/ (PLAN.md, SUMMARY.md, VERIFICATION.md, CONTEXT.md).planning/reports/SESSION_REPORT.md (last session outcomes)User input:
</context>
<process> Read and execute the forensics workflow from @${CLAUDE_PLUGIN_ROOT}/workflows/forensics.md end-to-end. </process>
<success_criteria>
.planning/forensics/report-{timestamp}.md</success_criteria>
<critical_rules>
</critical_rules>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | fail→fail | 14,270 | 7,442 | -48% | 1 | 1 | 0% | 2,480 | 1,674 | -33% | 0 | 0 | — |
case-01 | fail→fail | 16,684 | 5,115 | -69% | 1 | 1 | 0% | 2,887 | 785 | -73% | 0 | 0 | — |
case-02 | fail→fail | 3,784 | 19,460 | +414% | 1 | 1 | 0% | 173 | 987 | +471% | 0 | 0 | — |
case-03 | fail→fail | 15,665 | 2,888 | -82% | 1 | 1 | 0% | 2,589 | 747 | -71% | 0 | 0 | — |
case-04 | fail→fail | 4,443 | 6,291 | +42% | 1 | 1 | 0% | 731 | 1,072 | +47% | 0 | 0 | — |
case-05 | pass→fail | 16,345 | 24,672 | +51% | 1 | 1 | 0% | 3,139 | 1,240 | -60% | 0 | 0 | — |
case-06 | fail→fail | 2,312 | 4,355 | +88% | 1 | 1 | 0% | 412 | 1,230 | +199% | 0 | 0 | — |
case-07 | fail→fail | 4,076 | 3,664 | -10% | 1 | 1 | 0% | 647 | 1,139 | +76% | 0 | 0 | — |
case-08 | fail→fail | 10,875 | 5,426 | -50% | 1 | 1 | 0% | 1,930 | 1,092 | -43% | 0 | 0 | — |
case-14 | fail→pass | 9,469 | 2,941 | -69% | 1 | 1 | 0% | 1,458 | 965 | -34% | 0 | 0 | — |
case-09 | fail→fail | 14,865 | 6,969 | -53% | 1 | 1 | 0% | 2,472 | 893 | -64% | 0 | 0 | — |
case-10 | fail→fail | 2,432 | 4,941 | +103% | 1 | 1 | 0% | 296 | 842 | +184% | 0 | 0 | — |
case-11 | pass→fail | 13,736 | 4,461 | -68% | 1 | 1 | 0% | 2,504 | 766 | -69% | 0 | 0 | — |
case-12 | pass→fail | 2,724 | 5,022 | +84% | 1 | 1 | 0% | 450 | 735 | +63% | 0 | 0 | — |
case-13 | fail→fail | 4,022 | 5,584 | +39% | 1 | 1 | 0% | 231 | 954 | +313% | 0 | 0 | — |
case-16 | fail→pass | 15,680 | 21,537 | +37% | 1 | 1 | 0% | 2,806 | 4,467 | +59% | 0 | 0 | — |
case-17 | fail→fail | 2,930 | 7,181 | +145% | 1 | 1 | 0% | 441 | 793 | +80% | 0 | 0 | — |
case-18 | fail→fail | 24,723 | 4,825 | -80% | 1 | 1 | 0% | 234 | 766 | +227% | 0 | 0 | — |
case-19 | fail→fail | 4,441 | 5,010 | +13% | 1 | 1 | 0% | 220 | 742 | +237% | 0 | 0 | — |
case-20 | pass→fail | 8,747 | 4,959 | -43% | 1 | 1 | 0% | 1,371 | 751 | -45% | 0 | 0 | — |
case-21 | pass→fail | 19,732 | 6,282 | -68% | 1 | 1 | 0% | 3,274 | 693 | -79% | 0 | 0 | — |
case-22 | fail→fail | 2,637 | 4,595 | +74% | 1 | 1 | 0% | 385 | 709 | +84% | 0 | 0 | — |
case-23 | fail→fail | 4,582 | 5,432 | +19% | 1 | 1 | 0% | 253 | 811 | +221% | 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. 23 cases were attempted, and 5 counted toward the lift figure. The other 18 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 -13 percentage points is the difference between those two pass rates over the 5 comparable cases. 9 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.