Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →GitHub Wiki command center -- create, edit, organize, and search wiki pages entirely from the editor. Bypasses the drag-to-reorder, inconsistent navigation, and poorly-announced editor mode switches that make the wiki UI difficult for screen reader users.
.claude/skills/community-access-wiki-manager/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 613% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 119% | 0% |
Derived from .claude/agents/wiki-manager.md. Treat platform-specific tool names or delegation instructions as Codex equivalents.
Shared instructions
Skills: github-workflow-standards, github-scanning
You are the Wiki Manager. You give screen reader users and keyboard-only users full control over GitHub Wiki pages — a feature whose web UI relies on drag-to-reorder sidebars, inconsistent navigation landmarks, and editor mode switches that do not announce state changes to assistive technology.
You replace all of that with structured, navigable text output and simple git-based commands.
GitHub Wiki UI presents significant accessibility barriers:
This agent bypasses all of that by cloning the wiki git repository and working with pages as local markdown files.
_Sidebar.md with structured table of contents._Footer.md.GitHub wikis are backed by a git repository at {repo}.wiki.git. This agent clones the wiki repo, reads/creates/edits markdown files directly, then commits and pushes to publish.
gh api user.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 9,144 | 4,274 | -53% | 1 | 1 | 0% | 1,734 | 1,328 | -23% | 0 | 0 | — |
case-02 | fail→fail | 6,002 | 5,785 | -4% | 1 | 1 | 0% | 289 | 1,181 | +309% | 0 | 0 | — |
case-03 | fail→pass | 6,225 | 3,692 | -41% | 1 | 1 | 0% | 193 | 1,376 | +613% | 0 | 0 | — |
case-04 | fail→pass | 10,874 | 14,340 | +32% | 1 | 1 | 0% | 1,852 | 2,617 | +41% | 0 | 0 | — |
case-05 | fail→pass | 7,022 | 4,787 | -32% | 1 | 1 | 0% | 1,147 | 1,538 | +34% | 0 | 0 | — |
case-06 | pass→pass | 6,870 | 3,271 | -52% | 1 | 1 | 0% | 1,238 | 1,357 | +10% | 0 | 0 | — |
case-07 | pass→pass | 2,731 | 1,596 | -42% | 1 | 1 | 0% | 400 | 1,020 | +155% | 0 | 0 | — |
case-08 | pass→pass | 4,783 | 2,324 | -51% | 1 | 1 | 0% | 770 | 1,198 | +56% | 0 | 0 | — |
case-09 | pass→pass | 10,578 | 2,953 | -72% | 1 | 1 | 0% | 1,883 | 1,249 | -34% | 0 | 0 | — |
case-10 | fail→pass | 5,200 | 7,225 | +39% | 1 | 1 | 0% | 897 | 1,966 | +119% | 0 | 0 | — |
case-11 | pass→pass | 6,905 | 11,903 | +72% | 1 | 1 | 0% | 1,131 | 2,246 | +99% | 0 | 0 | — |
case-12 | fail→pass | 5,560 | 7,121 | +28% | 1 | 1 | 0% | 817 | 2,135 | +161% | 0 | 0 | — |
case-13 | fail→pass | 9,298 | 8,159 | -12% | 1 | 1 | 0% | 1,491 | 2,213 | +48% | 0 | 0 | — |
case-14 | pass→pass | 18,937 | 13,784 | -27% | 1 | 1 | 0% | 2,726 | 2,674 | -2% | 0 | 0 | — |
case-15 | pass→pass | 10,348 | 9,782 | -5% | 1 | 1 | 0% | 1,699 | 2,751 | +62% | 0 | 0 | — |
case-16 | pass→pass | 14,960 | 14,392 | -4% | 1 | 1 | 0% | 2,313 | 2,826 | +22% | 0 | 0 | — |
case-17 | pass→pass | 14,010 | 21,384 | +53% | 1 | 1 | 0% | 2,607 | 4,696 | +80% | 0 | 0 | — |
case-18 | pass→fail | 5,446 | 4,094 | -25% | 1 | 1 | 0% | 894 | 1,298 | +45% | 0 | 0 | — |
case-19 | pass→fail | 9,969 | 7,192 | -28% | 1 | 1 | 0% | 1,669 | 1,204 | -28% | 0 | 0 | — |
case-20 | pass→fail | 4,728 | 6,012 | +27% | 1 | 1 | 0% | 802 | 1,055 | +32% | 0 | 0 | — |
case-21 | pass→pass | 9,556 | 9,934 | +4% | 1 | 1 | 0% | 1,598 | 2,375 | +49% | 0 | 0 | — |
case-22 | fail→pass | 7,781 | 8,201 | +5% | 1 | 1 | 0% | 1,412 | 2,151 | +52% | 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 18 counted toward the lift figure. The other 4 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 +23 percentage points is the difference between those two pass rates over the 18 comparable cases. 3 cases got worse with the skill loaded, and they are 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.