Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Read this skill before updating changelogs
.claude/skills/dicklesworthstone-update-changelog/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -3% | 0% |
Update the repository changelog with changes between the last release and the current version (main) that are not yet incorporated. If CHANGELOG.md does not exist, use CHANGELOG instead.
If no baseline version is provided, use the most recent git tag. You can find it with git describe --tags --abbrev=0.
Use the following commands to gather commit information:
bash# Get the baseline version (if not provided) git describe --tags --abbrev=0 # Get all commits since the baseline version git log <baseline-version>..HEAD
Read the existing changelog file (CHANGELOG.md, or CHANGELOG if missing) and check if there are changes not yet incorporated, then add them. Always add them to the "Unreleased" section only. If there is none yet, add it at the top in the same style as the existing changelog (for example, ## Unreleased vs ## [Unreleased]).
#NUMBER) when available, but not raw commit hashes* or -) for individual changes foo.cleanup )markdown## 2.13.0 * Added multi-key support to the `|sort` filter. #827 * Fix `not undefined` with strict undefined behavior. #838 * Added support for free threading Python. #841 ## 2.12.0 * Item or attribute lookup will no longer swallow all errors in Python. #814 * Added `|zip` filter. #818 * Fix `break_on_hyphens` for the `|wordwrap` filter. #823 * Prefer error message from `unknown_method_callback`. #824 * Ignore `.jinja` and `.jinja2` as extensions in auto escape. #832
Good:
Fixed an issue with the TypeScript SDK which caused an incorrect config for CJS.Added support for claim timeout extension on checkpoint writes.Improved error reporting when task claim expires.Bad:
Fixed bug (too vague)Updated dependencies (insignificant unless it fixes a security issue)Refactored internal code structure (internal change, not user-facing)Fixed typo in comment (insignificant)main| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,592 | 4,967 | +38% | 1 | 1 | 0% | 597 | 977 | +64% | 0 | 0 | — |
case-02 | fail→fail | 3,971 | 4,481 | +13% | 1 | 1 | 0% | 520 | 1,009 | +94% | 0 | 0 | — |
case-03 | fail→fail | 6,920 | 4,155 | -40% | 1 | 1 | 0% | 312 | 995 | +219% | 0 | 0 | — |
case-04 | pass→pass | 8,885 | 3,546 | -60% | 1 | 1 | 0% | 1,445 | 1,324 | -8% | 0 | 0 | — |
case-05 | fail→pass | 11,362 | 8,419 | -26% | 1 | 1 | 0% | 1,813 | 1,699 | -6% | 0 | 0 | — |
case-06 | fail→pass | 8,355 | 2,058 | -75% | 1 | 1 | 0% | 1,395 | 1,122 | -20% | 0 | 0 | — |
case-07 | fail→pass | 11,414 | 3,323 | -71% | 1 | 1 | 0% | 1,953 | 1,371 | -30% | 0 | 0 | — |
case-08 | pass→pass | 6,846 | 4,014 | -41% | 1 | 1 | 0% | 1,262 | 1,413 | +12% | 0 | 0 | — |
case-09 | pass→pass | 7,262 | 2,676 | -63% | 1 | 1 | 0% | 1,301 | 1,263 | -3% | 0 | 0 | — |
case-10 | pass→pass | 10,245 | 8,612 | -16% | 1 | 1 | 0% | 1,702 | 2,315 | +36% | 0 | 0 | — |
case-11 | pass→pass | 7,134 | 3,614 | -49% | 1 | 1 | 0% | 1,304 | 1,489 | +14% | 0 | 0 | — |
case-12 | fail→pass | 10,296 | 4,006 | -61% | 1 | 1 | 0% | 1,555 | 1,383 | -11% | 0 | 0 | — |
case-13 | pass→pass | 7,950 | 2,458 | -69% | 1 | 1 | 0% | 1,270 | 1,128 | -11% | 0 | 0 | — |
case-19 | pass→pass | 4,297 | 3,909 | -9% | 1 | 1 | 0% | 639 | 1,411 | +121% | 0 | 0 | — |
case-14 | pass→pass | 8,137 | 4,424 | -46% | 1 | 1 | 0% | 1,468 | 1,395 | -5% | 0 | 0 | — |
case-15 | pass→pass | 12,415 | 11,810 | -5% | 1 | 1 | 0% | 2,014 | 2,319 | +15% | 0 | 0 | — |
case-16 | pass→pass | 11,077 | 1,570 | -86% | 1 | 1 | 0% | 1,667 | 1,050 | -37% | 0 | 0 | — |
case-17 | pass→pass | 11,969 | 6,831 | -43% | 1 | 1 | 0% | 1,868 | 1,226 | -34% | 0 | 0 | — |
case-18 | pass→pass | 10,931 | 2,338 | -79% | 1 | 1 | 0% | 1,600 | 1,186 | -26% | 0 | 0 | — |
case-20 | pass→pass | 9,162 | 2,740 | -70% | 1 | 1 | 0% | 1,377 | 1,206 | -12% | 0 | 0 | — |
case-21 | pass→pass | 10,500 | 5,221 | -50% | 1 | 1 | 0% | 1,506 | 1,577 | +5% | 0 | 0 | — |
case-22 | pass→fail | 8,642 | 4,143 | -52% | 1 | 1 | 0% | 1,507 | 1,458 | -3% | 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 19 counted toward the lift figure. The other 3 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 +14 percentage points is the difference between those two pass rates over the 19 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.