Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Interview the user and create an approved GRACE 4 GraceChangeSpec plus optional design-context.xml inside .grace/changes/active/C-*/.
.claude/skills/osovv-grace-spec/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -61% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -48% | 0% |
<skill> <change_bundle_contract> .grace/changes/active/C-CHANGE-ID/
spec.xml — normative GraceChangeSpecdesign-context.xml — optional, explanatory onlyplan.xml — created later by grace-plan</change_bundle_contract>
<status_rules> Create spec.xml as status="draft". Set status="approved" only after explicit user approval. Rejected or cancelled specs move to archive with terminal status. Do not create or edit plan.xml in this skill. </status_rules>
<strict_contract> The direct C-* wrapper must contain exactly one meaningful Summary, Goals, Constraints, NonGoals, AcceptanceCriteria, AffectedAreas, and VerificationIntent section. Empty containers are not approval-ready. Semantic anchors are canonical attribute-free XML tags, never attributes or attribute values. </strict_contract>
<workflow>
C-* change id.spec.xml from references/change-spec-template.xml with exactly one direct C-* wrapper and no empty required section.design-context.xml from its template.spec.xml as status="draft" and report the approval step needed.</workflow>
<hard_rules>
spec.xml is the source of truth for grace-plan; design context never adds requirements.grace lint --path <project-root> --assertions current as a pre-implementation active-baseline check after writing the bundle; never present it as target or final evidence.</hard_rules> </skill>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 12,567 | 12,532 | -0% | 1 | 1 | 0% | 2,146 | 956 | -55% | 0 | 0 | — |
case-02 | fail→fail | 18,878 | 2,720 | -86% | 1 | 1 | 0% | 3,171 | 974 | -69% | 0 | 0 | — |
case-03 | fail→fail | 6,757 | 2,700 | -60% | 1 | 1 | 0% | 1,227 | 875 | -29% | 0 | 0 | — |
case-04 | fail→pass | 11,576 | 13,007 | +12% | 1 | 1 | 0% | 1,897 | 1,325 | -30% | 0 | 0 | — |
case-17 | fail→pass | 4,570 | 2,695 | -41% | 1 | 1 | 0% | 780 | 982 | +26% | 0 | 0 | — |
case-05 | pass→pass | 4,740 | 3,648 | -23% | 1 | 1 | 0% | 840 | 1,211 | +44% | 0 | 0 | — |
case-06 | fail→pass | 8,520 | 2,944 | -65% | 1 | 1 | 0% | 1,501 | 1,006 | -33% | 0 | 0 | — |
case-07 | pass→pass | 7,366 | 2,340 | -68% | 1 | 1 | 0% | 1,296 | 993 | -23% | 0 | 0 | — |
case-08 | fail→pass | 21,743 | 4,378 | -80% | 1 | 1 | 0% | 3,349 | 1,303 | -61% | 0 | 0 | — |
case-09 | fail→pass | 11,231 | 2,969 | -74% | 1 | 1 | 0% | 1,936 | 1,013 | -48% | 0 | 0 | — |
case-10 | pass→pass | 9,595 | 4,184 | -56% | 1 | 1 | 0% | 1,747 | 1,330 | -24% | 0 | 0 | — |
case-11 | pass→pass | 5,866 | 1,760 | -70% | 1 | 1 | 0% | 1,006 | 842 | -16% | 0 | 0 | — |
case-12 | fail→pass | 9,273 | 4,399 | -53% | 1 | 1 | 0% | 1,463 | 1,153 | -21% | 0 | 0 | — |
case-13 | fail→pass | 12,709 | 6,371 | -50% | 1 | 1 | 0% | 2,767 | 1,562 | -44% | 0 | 0 | — |
case-14 | pass→fail | 5,934 | 7,104 | +20% | 1 | 1 | 0% | 1,010 | 1,816 | +80% | 0 | 0 | — |
case-15 | pass→pass | 9,360 | 3,619 | -61% | 1 | 1 | 0% | 1,550 | 1,080 | -30% | 0 | 0 | — |
case-16 | pass→pass | 9,114 | 2,718 | -70% | 1 | 1 | 0% | 1,496 | 1,022 | -32% | 0 | 0 | — |
case-18 | pass→pass | 9,608 | 2,495 | -74% | 1 | 1 | 0% | 1,602 | 862 | -46% | 0 | 0 | — |
case-19 | pass→pass | 7,800 | 2,294 | -71% | 1 | 1 | 0% | 1,288 | 883 | -31% | 0 | 0 | — |
case-20 | fail→pass | 20,674 | 3,484 | -83% | 1 | 1 | 0% | 3,994 | 1,092 | -73% | 0 | 0 | — |
case-21 | fail→pass | 1,862 | 3,289 | +77% | 1 | 1 | 0% | 235 | 1,114 | +374% | 0 | 0 | — |
case-22 | fail→pass | 18,680 | 6,555 | -65% | 1 | 1 | 0% | 3,493 | 1,446 | -59% | 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. The headline lift of +41 percentage points is the difference between those two pass rates over the 22 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.