Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Implement and test Go codemods for the gh aw fix command.
.claude/skills/github-go-codemod/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -49% | 0% |
Use this skill when adding or updating codemods used by gh aw fix.
pkg/cli/fix_command.go to understand execution flow:GetAllCodemods()(newContent, applied, error) and be safe for no-op inputpkg/cli/fix_codemods.go.pkg/cli/yaml_frontmatter_utils.go and reusable codemod helper constructors in pkg/cli/codemod_factory.go.pkg/cli/ named codemod_<feature>.go.logger.New("cli:codemod_<feature>").get<Feature>Codemod() Codemod and populate all metadata fields:ID (stable, unique)Name (human-readable)Description (clear migration behavior)IntroducedIn (release version)Apply:applied=false when migration is not neededapplyFrontmatterLineTransform from pkg/cli/yaml_frontmatter_utils.gofindAndReplaceInLineremoveFieldFromBlockremoveParentBlockIfTrulyEmptynewFieldRemovalCodemodnewMoveTopLevelKeyToOnBlockCodemod...WithDeps constructor so tests can mock behavior.pkg/cli/fix_codemods.go within GetAllCodemods().Create pkg/cli/codemod_<feature>_test.go and cover:
ID, Name, Description, IntroducedIn, Apply != nil)....WithDeps exists (success and fallback/error paths).When complexity is high (expression rewriting, parser-like behavior), add fuzz tests in codemod_<feature>_fuzz_test.go using the pattern in pkg/cli/codemod_steps_run_secrets_env_fuzz_test.go.
After adding a codemod, update pkg/cli/fix_codemods_test.go:
Order matters because codemods run sequentially and later codemods observe prior transformations.
Run targeted checks first:
go test -v ./pkg/cli -run Codemod -count=1go test -v ./pkg/cli -run Fix -count=1Then run repository standards:
make buildmake test-unitmake lintA codemod is ready only when it is:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 20,157 | 5,757 | -71% | 1 | 1 | 0% | 3,993 | 1,235 | -69% | 0 | 0 | — |
case-02 | fail→pass | 12,421 | 8,426 | -32% | 1 | 1 | 0% | 2,376 | 2,562 | +8% | 0 | 0 | — |
case-03 | fail→fail | 20,822 | 16,307 | -22% | 1 | 1 | 0% | 3,229 | 4,600 | +42% | 0 | 0 | — |
case-04 | fail→pass | 11,214 | 2,788 | -75% | 1 | 1 | 0% | 1,797 | 1,358 | -24% | 0 | 0 | — |
case-05 | fail→pass | 11,683 | 3,459 | -70% | 1 | 1 | 0% | 2,267 | 1,480 | -35% | 0 | 0 | — |
case-06 | fail→pass | 7,828 | 3,173 | -59% | 1 | 1 | 0% | 1,183 | 1,388 | +17% | 0 | 0 | — |
case-07 | fail→pass | 17,289 | 5,699 | -67% | 1 | 1 | 0% | 2,799 | 1,427 | -49% | 0 | 0 | — |
case-08 | fail→pass | 8,207 | 2,926 | -64% | 1 | 1 | 0% | 1,509 | 1,299 | -14% | 0 | 0 | — |
case-09 | fail→pass | 8,531 | 2,573 | -70% | 1 | 1 | 0% | 1,325 | 1,419 | +7% | 0 | 0 | — |
case-10 | pass→fail | 7,543 | 1,387 | -82% | 1 | 1 | 0% | 1,339 | 1,184 | -12% | 0 | 0 | — |
case-11 | fail→fail | 9,595 | 2,223 | -77% | 1 | 1 | 0% | 1,484 | 1,223 | -18% | 0 | 0 | — |
case-12 | fail→pass | 10,331 | 1,831 | -82% | 1 | 1 | 0% | 1,596 | 1,146 | -28% | 0 | 0 | — |
case-13 | fail→pass | 6,600 | 2,045 | -69% | 1 | 1 | 0% | 945 | 1,178 | +25% | 0 | 0 | — |
case-14 | fail→pass | 12,811 | 3,305 | -74% | 1 | 1 | 0% | 2,216 | 1,404 | -37% | 0 | 0 | — |
case-15 | fail→pass | 8,072 | 3,741 | -54% | 1 | 1 | 0% | 1,351 | 1,474 | +9% | 0 | 0 | — |
case-16 | fail→pass | 14,227 | 3,856 | -73% | 1 | 1 | 0% | 2,334 | 1,596 | -32% | 0 | 0 | — |
case-17 | fail→pass | 10,939 | 3,493 | -68% | 1 | 1 | 0% | 1,752 | 1,509 | -14% | 0 | 0 | — |
case-18 | fail→pass | 11,129 | 2,768 | -75% | 1 | 1 | 0% | 1,991 | 1,387 | -30% | 0 | 0 | — |
case-19 | fail→pass | 12,569 | 3,019 | -76% | 1 | 1 | 0% | 2,175 | 1,378 | -37% | 0 | 0 | — |
case-20 | pass→pass | 10,863 | 8,560 | -21% | 1 | 1 | 0% | 2,051 | 2,433 | +19% | 0 | 0 | — |
case-21 | pass→pass | 15,829 | 13,262 | -16% | 1 | 1 | 0% | 3,113 | 3,286 | +6% | 0 | 0 | — |
case-22 | pass→pass | 20,390 | 12,533 | -39% | 1 | 1 | 0% | 3,811 | 3,272 | -14% | 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 21 counted toward the lift figure. The other 1 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 +64 percentage points is the difference between those two pass rates over the 21 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.