Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Agent-applied GRACE 3 to GRACE 4 migration workflow. CLI validates the result but does not convert or delete files.
.claude/skills/osovv-grace-migrate/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -22% | 0% |
<skill> <migration_safety> Migration is copy-and-validate, not destructive conversion. Before writing .grace, inventory every legacy source, record a restorable backup path outside the proposed cleanup set, present the migration report, and wait for explicit write approval. Never create retroactive C-* bundles. </migration_safety>
<workflow>
.grace state. If .grace already exists, stop and ask whether migration should merge, replace, or abort.references/migration-report-template.xml, review references/migration-checklist.md, and ask for explicit approval before writing .grace..grace, then run grace lint --path <project-root> --assertions current and grace status --path <project-root> --json. Require status projectKind grace4 with no integrity errors, and record commands, exit states, and findings.git status --porcelain --untracked-files=all before cleanup.</workflow>
<cleanup_preconditions> Cleanup requires all of: complete inventory, restorable backup, explicit write approval, successful current lint, fresh status proving GRACE 4 with no integrity errors, verified generated coverage, git availability/worktree inspection, recorded porcelain output when applicable, exact proposed paths, separate explicit cleanup approval, and separate dirty or non-git risk acknowledgement when applicable. A missing gate means no cleanup. </cleanup_preconditions>
<cleanup_failure_rule> If validation, git inspection, move, archive, or delete fails, stop immediately and state: "Legacy GRACE 3 artifacts remain untouched unless the failure output explicitly lists a completed move." Never retry destructive cleanup automatically. </cleanup_failure_rule> </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 | 12,186 | 9,654 | -21% | 1 | 1 | 0% | 2,468 | 835 | -66% | 0 | 0 | — |
case-02 | fail→fail | 12,494 | 2,508 | -80% | 1 | 1 | 0% | 2,436 | 965 | -60% | 0 | 0 | — |
case-03 | fail→fail | 15,060 | 2,566 | -83% | 1 | 1 | 0% | 3,068 | 955 | -69% | 0 | 0 | — |
case-04 | fail→pass | 8,840 | 3,648 | -59% | 1 | 1 | 0% | 1,496 | 1,228 | -18% | 0 | 0 | — |
case-05 | pass→pass | 14,603 | 3,634 | -75% | 1 | 1 | 0% | 2,049 | 1,233 | -40% | 0 | 0 | — |
case-06 | fail→pass | 10,239 | 4,176 | -59% | 1 | 1 | 0% | 1,674 | 1,272 | -24% | 0 | 0 | — |
case-07 | fail→pass | 11,284 | 1,937 | -83% | 1 | 1 | 0% | 1,965 | 934 | -52% | 0 | 0 | — |
case-08 | fail→pass | 8,707 | 4,637 | -47% | 1 | 1 | 0% | 1,521 | 1,422 | -7% | 0 | 0 | — |
case-09 | fail→pass | 7,444 | 2,332 | -69% | 1 | 1 | 0% | 1,238 | 969 | -22% | 0 | 0 | — |
case-10 | pass→pass | 9,216 | 4,886 | -47% | 1 | 1 | 0% | 1,550 | 1,501 | -3% | 0 | 0 | — |
case-11 | pass→pass | 9,248 | 4,309 | -53% | 1 | 1 | 0% | 1,589 | 1,394 | -12% | 0 | 0 | — |
case-12 | fail→fail | 11,916 | 4,282 | -64% | 1 | 1 | 0% | 1,897 | 1,333 | -30% | 0 | 0 | — |
case-13 | fail→pass | 9,997 | 2,360 | -76% | 1 | 1 | 0% | 1,646 | 950 | -42% | 0 | 0 | — |
case-14 | pass→pass | 10,288 | 5,634 | -45% | 1 | 1 | 0% | 1,752 | 1,480 | -16% | 0 | 0 | — |
case-15 | fail→pass | 9,789 | 5,070 | -48% | 1 | 1 | 0% | 1,687 | 1,531 | -9% | 0 | 0 | — |
case-16 | pass→pass | 9,218 | 4,012 | -56% | 1 | 1 | 0% | 1,572 | 1,291 | -18% | 0 | 0 | — |
case-17 | pass→pass | 5,478 | 3,730 | -32% | 1 | 1 | 0% | 873 | 1,271 | +46% | 0 | 0 | — |
case-18 | pass→pass | 8,568 | 3,477 | -59% | 1 | 1 | 0% | 1,402 | 1,226 | -13% | 0 | 0 | — |
case-19 | pass→pass | 13,128 | 7,489 | -43% | 1 | 1 | 0% | 2,541 | 1,958 | -23% | 0 | 0 | — |
case-20 | pass→pass | 8,175 | 7,029 | -14% | 1 | 1 | 0% | 1,592 | 1,854 | +16% | 0 | 0 | — |
case-21 | pass→pass | 15,302 | 11,296 | -26% | 1 | 1 | 0% | 2,575 | 2,382 | -7% | 0 | 0 | — |
case-22 | pass→pass | 10,208 | 3,791 | -63% | 1 | 1 | 0% | 1,701 | 1,211 | -29% | 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 +32 percentage points is the difference between those two pass rates over the 22 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.