Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Add a new changelog entry to docs/changelog-entries/
.claude/skills/elie222-changelog/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 113% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 113% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -1% | 0% |
Add changelog entries as individual files in docs/changelog-entries/, then regenerate docs/changelog.mdx before opening or updating the PR.
Use a single long-lived branch for changelog automation: automation/changelog. The automation should update the existing open changelog PR from that branch when possible instead of opening multiple concurrent changelog PRs.
description field (e.g., "Chat Everywhere", not "v2.28"). The theme should immediately tell users what changed.Create or update docs/changelog-entries/YYYY-MM-DD.mdx with frontmatter + markdown:
mdx--- description: "Headline Theme" --- One or two sentences about the main feature. - Bullet one - Bullet two - Bullet three
The date is derived from the filename automatically.
Do not hand-edit docs/changelog.mdx. Regenerate it with node docs/scripts/build-changelog.mjs.
gh pr list --repo elie222/inbox-zero --state merged --limit 30 --json number,title,mergedAtautomation/changelog to main: gh pr list --repo elie222/inbox-zero --head automation/changelog --base main --state open --json number,title,urldocs/changelog-entries/YYYY-MM-DD.mdx with frontmatter (description) and markdown contentdocs/changelog.mdx: node docs/scripts/build-changelog.mjsdocs/changelog-entries/YYYY-MM-DD.mdx and docs/changelog.mdxautomation/changelog exists, update that branch and PR; otherwise create it from automation/changelogBefore making changes, reset the automation branch to the latest main so each run starts from a clean base:
bashgit fetch origin git checkout -B automation/changelog origin/main
After updating the changelog files, push that branch and create or update the PR from automation/changelog to main.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 4,344 | 6,189 | +42% | 1 | 1 | 0% | 228 | 1,341 | +488% | 0 | 0 | — |
case-01 | fail→fail | 4,517 | 13,515 | +199% | 1 | 1 | 0% | 204 | 1,179 | +478% | 0 | 0 | — |
case-03 | fail→fail | 6,185 | 6,084 | -2% | 1 | 1 | 0% | 226 | 1,322 | +485% | 0 | 0 | — |
case-04 | pass→pass | 11,541 | 10,123 | -12% | 1 | 1 | 0% | 1,912 | 2,526 | +32% | 0 | 0 | — |
case-05 | pass→pass | 5,408 | 5,712 | +6% | 1 | 1 | 0% | 977 | 1,907 | +95% | 0 | 0 | — |
case-06 | pass→pass | 13,707 | 19,564 | +43% | 1 | 1 | 0% | 3,071 | 5,127 | +67% | 0 | 0 | — |
case-07 | fail→pass | 5,953 | 8,375 | +41% | 1 | 1 | 0% | 1,164 | 2,480 | +113% | 0 | 0 | — |
case-08 | fail→pass | 7,119 | 10,384 | +46% | 1 | 1 | 0% | 1,295 | 2,753 | +113% | 0 | 0 | — |
case-09 | fail→pass | 9,429 | 6,745 | -28% | 1 | 1 | 0% | 1,470 | 2,070 | +41% | 0 | 0 | — |
case-10 | pass→pass | 10,267 | 4,345 | -58% | 1 | 1 | 0% | 1,799 | 1,713 | -5% | 0 | 0 | — |
case-11 | fail→pass | 11,019 | 3,327 | -70% | 1 | 1 | 0% | 2,033 | 1,468 | -28% | 0 | 0 | — |
case-12 | fail→pass | 10,469 | 4,098 | -61% | 1 | 1 | 0% | 1,554 | 1,546 | -1% | 0 | 0 | — |
case-13 | pass→pass | 9,078 | 2,404 | -74% | 1 | 1 | 0% | 1,365 | 1,224 | -10% | 0 | 0 | — |
case-14 | pass→pass | 8,064 | 1,995 | -75% | 1 | 1 | 0% | 1,227 | 1,146 | -7% | 0 | 0 | — |
case-15 | pass→pass | 8,101 | 2,502 | -69% | 1 | 1 | 0% | 1,373 | 1,341 | -2% | 0 | 0 | — |
case-16 | fail→pass | 10,656 | 5,137 | -52% | 1 | 1 | 0% | 1,845 | 1,829 | -1% | 0 | 0 | — |
case-17 | fail→pass | 8,566 | 2,558 | -70% | 1 | 1 | 0% | 1,494 | 1,270 | -15% | 0 | 0 | — |
case-18 | fail→pass | 7,297 | 2,146 | -71% | 1 | 1 | 0% | 1,224 | 1,205 | -2% | 0 | 0 | — |
case-19 | pass→pass | 5,492 | 1,671 | -70% | 1 | 1 | 0% | 975 | 1,137 | +17% | 0 | 0 | — |
case-20 | fail→pass | 8,254 | 2,634 | -68% | 1 | 1 | 0% | 1,446 | 1,300 | -10% | 0 | 0 | — |
case-21 | pass→pass | 8,501 | 3,406 | -60% | 1 | 1 | 0% | 1,222 | 1,385 | +13% | 0 | 0 | — |
case-22 | fail→pass | 8,367 | 3,709 | -56% | 1 | 1 | 0% | 1,501 | 1,563 | +4% | 0 | 0 | — |
case-23 | pass→pass | 6,711 | 2,309 | -66% | 1 | 1 | 0% | 917 | 1,282 | +40% | 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. 23 cases were attempted, and 20 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 +43 percentage points is the difference between those two pass rates over the 20 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.