Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Choose an implementation path (direct or plan) for evaluated findings and dispatch it. Direct path applies fixes directly; plan path runs /turboplan. Use after /evaluate-findings has tagged findings and they need to be implemented, or when the user asks to "resolve findings", "apply evaluated findings", or "dispatch findings to implementation".
.claude/skills/tobihagemann-resolve-findings/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -55% | 0% |
Choose a path for evaluated findings and run it. Direct path applies fixes directly; plan path runs $turboplan.
At the start, use update_plan to track each step, restating any remaining steps of a parent workflow alongside them:
Present a summary of accepted findings (Apply verdict): count by complexity (mechanical fixes vs. architectural or design changes). Then use request_user_input to let the user choose:
$turboplan for drafting, refinement, approval, implementation, and finalize$apply-findings, then $finalizeSuggest Plan when findings include complex or architectural changes. Suggest Direct when all findings are mechanical fixes.
Add a Get a second opinion option whenever the findings would establish a pattern others will follow, change an interface, or commit to a data shape, and whenever neither path earns the suggestion with conviction. It runs the $consult-claude skill for which path the findings warrant on technical merit. Then resolve the path with that answer in hand, re-asking when the choice stays the user's.
Read the reference file for the confirmed path and follow its phases:
State the chosen path before continuing with the reference file.
Then call update_plan to mark this step completed and continue with the next step of the active workflow.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→pass | 20,052 | 26,708 | +33% | 1 | 1 | 0% | 3,309 | 5,058 | +53% | 0 | 0 | — |
case-01 | fail→fail | 9,145 | 7,937 | -13% | 1 | 1 | 0% | 1,409 | 582 | -59% | 0 | 0 | — |
case-02 | fail→fail | 2,915 | 5,610 | +92% | 1 | 1 | 0% | 401 | 685 | +71% | 0 | 0 | — |
case-03 | fail→fail | 9,346 | 6,742 | -28% | 1 | 1 | 0% | 1,378 | 674 | -51% | 0 | 0 | — |
case-04 | fail→fail | 4,268 | 23,118 | +442% | 1 | 1 | 0% | 597 | 2,342 | +292% | 0 | 0 | — |
case-06 | fail→fail | 9,030 | 9,180 | +2% | 1 | 1 | 0% | 1,494 | 977 | -35% | 0 | 0 | — |
case-07 | pass→pass | 13,106 | 4,726 | -64% | 1 | 1 | 0% | 2,076 | 1,126 | -46% | 0 | 0 | — |
case-08 | pass→pass | 7,538 | 2,402 | -68% | 1 | 1 | 0% | 1,149 | 776 | -32% | 0 | 0 | — |
case-09 | fail→fail | 14,422 | 14,306 | -1% | 1 | 1 | 0% | 2,192 | 1,632 | -26% | 0 | 0 | — |
case-10 | fail→pass | 12,734 | 3,543 | -72% | 1 | 1 | 0% | 1,967 | 988 | -50% | 0 | 0 | — |
case-11 | pass→pass | 16,711 | 4,446 | -73% | 1 | 1 | 0% | 2,614 | 1,144 | -56% | 0 | 0 | — |
case-12 | fail→pass | 10,274 | 2,317 | -77% | 1 | 1 | 0% | 1,582 | 762 | -52% | 0 | 0 | — |
case-13 | fail→pass | 10,255 | 1,747 | -83% | 1 | 1 | 0% | 1,589 | 636 | -60% | 0 | 0 | — |
case-14 | fail→fail | 8,797 | 1,371 | -84% | 1 | 1 | 0% | 1,310 | 574 | -56% | 0 | 0 | — |
case-15 | fail→pass | 12,239 | 3,548 | -71% | 1 | 1 | 0% | 1,967 | 946 | -52% | 0 | 0 | — |
case-16 | fail→pass | 9,633 | 2,101 | -78% | 1 | 1 | 0% | 1,519 | 686 | -55% | 0 | 0 | — |
case-17 | fail→pass | 9,463 | 1,943 | -79% | 1 | 1 | 0% | 1,436 | 640 | -55% | 0 | 0 | — |
case-18 | fail→fail | 7,254 | 1,959 | -73% | 1 | 1 | 0% | 1,102 | 666 | -40% | 0 | 0 | — |
case-19 | fail→fail | 10,222 | 2,804 | -73% | 1 | 1 | 0% | 1,564 | 812 | -48% | 0 | 0 | — |
case-20 | fail→fail | 11,266 | 5,660 | -50% | 1 | 1 | 0% | 1,757 | 1,235 | -30% | 0 | 0 | — |
case-21 | fail→pass | 14,446 | 6,204 | -57% | 1 | 1 | 0% | 2,128 | 1,334 | -37% | 0 | 0 | — |
case-22 | pass→pass | 10,657 | 3,105 | -71% | 1 | 1 | 0% | 1,689 | 870 | -48% | 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 17 counted toward the lift figure. The other 5 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 +32 percentage points is the difference between those two pass rates over the 17 comparable cases. 1 case got worse with the skill loaded, and it is 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.