Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Automatically hunt for high-impact OSS contribution opportunities in trending repositories.
.claude/skills/sickn33-oss-hunter/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -56% | 0% |
A precision skill for agents to find, analyze, and strategize for high-impact Open Source contributions. This skill helps you become a top-tier contributor by identifying the most "mergeable" and influential issues in trending repositories.
Ask your agent:
When hunting for contributions, the agent follows this multi-stage protocol:
Use web_search or gh api to find trending repositories. Focus on:
Search for specific labels:
help-wantedgood-first-issuebugv1 / roadmapbashgh issue list --repo owner/repo --label "help wanted" --limit 10
Analyze the issue:
Generate a structured report for the human:
gh CLI or web_search tools.Build a better hunter by adding new heuristics to Phase 3. Submit your improvements to the ClawForge.
Powered by OpenClaw & ClawForge.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 20,972 | 17,907 | -15% | 1 | 1 | 0% | 3,881 | 3,182 | -18% | 0 | 0 | — |
case-02 | fail→fail | 27,925 | 33,230 | +19% | 1 | 1 | 0% | 5,012 | 4,828 | -4% | 0 | 0 | — |
case-03 | fail→fail | 20,705 | 32,099 | +55% | 1 | 1 | 0% | 4,087 | 1,056 | -74% | 0 | 0 | — |
case-04 | fail→fail | 2,708 | 4,247 | +57% | 1 | 1 | 0% | 471 | 1,285 | +173% | 0 | 0 | — |
case-05 | pass→pass | 6,393 | 5,413 | -15% | 1 | 1 | 0% | 1,381 | 1,636 | +18% | 0 | 0 | — |
case-06 | fail→fail | 2,136 | 2,071 | -3% | 1 | 1 | 0% | 390 | 972 | +149% | 0 | 0 | — |
case-07 | fail→pass | 11,831 | 10,774 | -9% | 1 | 1 | 0% | 2,100 | 2,429 | +16% | 0 | 0 | — |
case-08 | fail→pass | 9,299 | 1,758 | -81% | 1 | 1 | 0% | 1,551 | 868 | -44% | 0 | 0 | — |
case-09 | pass→pass | 14,184 | 2,054 | -86% | 1 | 1 | 0% | 2,367 | 950 | -60% | 0 | 0 | — |
case-10 | pass→pass | 10,054 | 6,001 | -40% | 1 | 1 | 0% | 1,776 | 1,788 | +1% | 0 | 0 | — |
case-11 | fail→fail | 3,011 | 2,324 | -23% | 1 | 1 | 0% | 504 | 1,039 | +106% | 0 | 0 | — |
case-12 | pass→pass | 6,770 | 1,571 | -77% | 1 | 1 | 0% | 1,096 | 848 | -23% | 0 | 0 | — |
case-13 | pass→pass | 14,962 | 11,442 | -24% | 1 | 1 | 0% | 2,437 | 2,538 | +4% | 0 | 0 | — |
case-14 | pass→pass | 9,393 | 1,399 | -85% | 1 | 1 | 0% | 1,532 | 792 | -48% | 0 | 0 | — |
case-15 | fail→pass | 6,768 | 1,663 | -75% | 1 | 1 | 0% | 1,090 | 872 | -20% | 0 | 0 | — |
case-16 | fail→pass | 8,085 | 1,858 | -77% | 1 | 1 | 0% | 1,429 | 957 | -33% | 0 | 0 | — |
case-17 | pass→pass | 8,500 | 2,185 | -74% | 1 | 1 | 0% | 1,326 | 966 | -27% | 0 | 0 | — |
case-18 | pass→pass | 13,790 | 7,287 | -47% | 1 | 1 | 0% | 944 | 1,382 | +46% | 0 | 0 | — |
case-19 | pass→pass | 11,833 | 3,179 | -73% | 1 | 1 | 0% | 1,841 | 948 | -49% | 0 | 0 | — |
case-20 | fail→pass | 10,938 | 1,767 | -84% | 1 | 1 | 0% | 1,887 | 838 | -56% | 0 | 0 | — |
case-21 | pass→pass | 11,191 | 6,317 | -44% | 1 | 1 | 0% | 2,063 | 1,802 | -13% | 0 | 0 | — |
case-22 | fail→pass | 6,902 | 2,188 | -68% | 1 | 1 | 0% | 1,125 | 982 | -13% | 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 +27 percentage points is the difference between those two pass rates over the 21 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.