Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Sync skills between local installation and the GitHub source-of-truth repository. Use when asked to install, update, list, or push skills.
.claude/skills/jdrhyne-skill-sync/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 82% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -62% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -72% | 0% |
Manage skills from the shared GitHub repository with GitHub as canonical source of truth.
> Never commit secrets. Keep keys/tokens out of SKILL.md and scripts.
bash# List available skills in the repo skill-sync list # Install one skill skill-sync install <skill-name> # Install/update all skills from repo skill-sync install --all # Push a local skill update to repo via PR skill-sync push <skill-name> # Refresh local repo cache skill-sync update
skill-sync listShows all skills available in the remote repository.
skill-sync install <name>Installs/updates a skill from repo into local skills directory.
skill-sync install --allInstalls/updates all skills from repo.
skill-sync push <name>Pushes local skill changes via branch + PR (gh CLI).
skill-sync updatePulls latest repo changes without installing.
Default paths:
~/.agent-skills-repo~/clawd/skills (or $CLAWD_SKILLS_DIR)https://github.com/jdrhyne/agent-skills.git| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 2,539 | 10,571 | +316% | 1 | 1 | 0% | 344 | 627 | +82% | 0 | 0 | — |
case-02 | fail→pass | 7,708 | 2,581 | -67% | 1 | 1 | 0% | 1,144 | 640 | -44% | 0 | 0 | — |
case-03 | fail→pass | 5,050 | 1,634 | -68% | 1 | 1 | 0% | 775 | 500 | -35% | 0 | 0 | — |
case-04 | fail→pass | 8,771 | 1,684 | -81% | 1 | 1 | 0% | 1,299 | 488 | -62% | 0 | 0 | — |
case-05 | fail→pass | 14,752 | 1,909 | -87% | 1 | 1 | 0% | 2,106 | 590 | -72% | 0 | 0 | — |
case-06 | fail→pass | 9,952 | 1,553 | -84% | 1 | 1 | 0% | 1,460 | 491 | -66% | 0 | 0 | — |
case-07 | fail→pass | 16,474 | 2,216 | -87% | 1 | 1 | 0% | 2,478 | 622 | -75% | 0 | 0 | — |
case-08 | fail→pass | 11,885 | 1,754 | -85% | 1 | 1 | 0% | 1,985 | 577 | -71% | 0 | 0 | — |
case-09 | fail→pass | 14,680 | 1,707 | -88% | 1 | 1 | 0% | 2,272 | 515 | -77% | 0 | 0 | — |
case-10 | fail→pass | 10,602 | 2,860 | -73% | 1 | 1 | 0% | 1,768 | 739 | -58% | 0 | 0 | — |
case-11 | fail→pass | 13,050 | 1,858 | -86% | 1 | 1 | 0% | 1,995 | 545 | -73% | 0 | 0 | — |
case-12 | pass→pass | 11,688 | 3,873 | -67% | 1 | 1 | 0% | 1,813 | 924 | -49% | 0 | 0 | — |
case-13 | fail→pass | 10,994 | 1,738 | -84% | 1 | 1 | 0% | 1,525 | 555 | -64% | 0 | 0 | — |
case-14 | fail→pass | 16,215 | 2,217 | -86% | 1 | 1 | 0% | 1,045 | 602 | -42% | 0 | 0 | — |
case-15 | fail→pass | 9,660 | 1,799 | -81% | 1 | 1 | 0% | 1,501 | 554 | -63% | 0 | 0 | — |
case-16 | fail→pass | 7,577 | 1,721 | -77% | 1 | 1 | 0% | 1,271 | 575 | -55% | 0 | 0 | — |
case-17 | fail→pass | 6,101 | 2,793 | -54% | 1 | 1 | 0% | 922 | 742 | -20% | 0 | 0 | — |
case-18 | pass→pass | 9,298 | 1,800 | -81% | 1 | 1 | 0% | 1,359 | 500 | -63% | 0 | 0 | — |
case-19 | fail→pass | 15,134 | 1,820 | -88% | 1 | 1 | 0% | 2,290 | 522 | -77% | 0 | 0 | — |
case-20 | pass→pass | 8,707 | 6,324 | -27% | 1 | 1 | 0% | 1,403 | 1,340 | -4% | 0 | 0 | — |
case-21 | pass→pass | 5,098 | 3,509 | -31% | 1 | 1 | 0% | 820 | 879 | +7% | 0 | 0 | — |
case-22 | pass→pass | 6,203 | 4,372 | -30% | 1 | 1 | 0% | 1,074 | 975 | -9% | 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 +77 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.