Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Download and install Claude Code skills from various sources. Supports GitHub repositories, compressed archives (.zip, .tar.gz, .skill), and direct URLs. Use when user wants to download, install, or add a skill from GitHub, URL, or archive file. Triggers on "download skill", "install skill", "add skill from", "get skill".
.claude/skills/nicepkg-skill-downloader/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -36% | 0% |
Download and install Claude Code skills from multiple sources.
| Source Type | Examples | |-------------|----------| | GitHub Repository | https://github.com/user/repo, github.com/user/repo | | Compressed Archive | .zip, .tar.gz, .tgz, .skill (renamed zip) | | Direct URL | URL pointing to archive or skill folder |
bashpython scripts/download_from_github.py <repo-url> <skill-path> --output <target-dir>
Examples:
bash# Official Anthropic skills python scripts/download_from_github.py https://github.com/anthropics/skills skills/docx --output ./.claude/skills/ # Community skills (root level) python scripts/download_from_github.py https://github.com/gked2121/claude-skills social-repurposer --output ./.claude/skills/ # Nested skill path python scripts/download_from_github.py https://github.com/MadAppGang/claude-code skills/content-brief --output ./.claude/skills/
bashpython scripts/download_from_archive.py <url-or-path> --output <target-dir>
Examples:
bash# From URL python scripts/download_from_archive.py https://example.com/skills/my-skill.zip --output ./.claude/skills/ # From local file python scripts/download_from_archive.py ./downloads/my-skill.tar.gz --output ./.claude/skills/ # .skill files (renamed zip) python scripts/download_from_archive.py https://skillhub.club/download/awesome-skill.skill --output ./.claude/skills/
bashpython scripts/download_skill.py <source> --output <target-dir>
Automatically detects source type:
Examples:
bash# Auto-detect GitHub python scripts/download_skill.py https://github.com/anthropics/skills/tree/main/skills/docx --output ./.claude/skills/ # Auto-detect archive python scripts/download_skill.py https://example.com/my-skill.zip --output ./.claude/skills/
All download methods validate:
SKILL.mdSKILL.md has valid YAML frontmatter with name and descriptionDownloaded skills are placed in:
<output-dir>/
└── <skill-name>/
├── SKILL.md # Required
├── scripts/ # Optional
└── ...| Error | Cause | Solution | |-------|-------|----------| | Skill already exists | Target directory exists | Use --force to overwrite | | SKILL.md not found | Invalid skill package | Verify source contains SKILL.md | | Invalid archive | Corrupted or unsupported format | Check file integrity | | Network error | Download failed | Retry or check URL |
When used with workflow-creator, download skills to workflow's .claude/skills/ directory:
bash# Create workflow first python /path/to/workflow-creator/scripts/create_workflow.py my-workflow --path ./workflows # Then download skills python scripts/download_skill.py https://github.com/anthropics/skills/tree/main/skills/docx --output ./workflows/my-workflow/.claude/skills/
Popular skill repositories:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 8,603 | 23,907 | +178% | 1 | 1 | 0% | 1,663 | 1,884 | +13% | 0 | 0 | — |
case-02 | fail→pass | 6,745 | 9,857 | +46% | 1 | 1 | 0% | 1,162 | 2,121 | +83% | 0 | 0 | — |
case-03 | fail→pass | 8,137 | 3,867 | -52% | 1 | 1 | 0% | 1,199 | 1,860 | +55% | 0 | 0 | — |
case-04 | fail→pass | 7,927 | 2,902 | -63% | 1 | 1 | 0% | 1,256 | 1,627 | +30% | 0 | 0 | — |
case-05 | fail→pass | 16,000 | 2,090 | -87% | 1 | 1 | 0% | 2,384 | 1,535 | -36% | 0 | 0 | — |
case-06 | fail→pass | 11,527 | 3,981 | -65% | 1 | 1 | 0% | 1,855 | 1,714 | -8% | 0 | 0 | — |
case-12 | pass→pass | 5,872 | 4,034 | -31% | 1 | 1 | 0% | 1,005 | 1,677 | +67% | 0 | 0 | — |
case-07 | fail→pass | 20,457 | 3,081 | -85% | 1 | 1 | 0% | 4,318 | 1,713 | -60% | 0 | 0 | — |
case-08 | fail→pass | 6,305 | 2,847 | -55% | 1 | 1 | 0% | 981 | 1,623 | +65% | 0 | 0 | — |
case-09 | fail→pass | 12,249 | 3,471 | -72% | 1 | 1 | 0% | 1,811 | 1,723 | -5% | 0 | 0 | — |
case-10 | pass→pass | 13,378 | 6,242 | -53% | 1 | 1 | 0% | 2,321 | 2,006 | -14% | 0 | 0 | — |
case-11 | pass→pass | 10,182 | 4,004 | -61% | 1 | 1 | 0% | 1,479 | 1,580 | +7% | 0 | 0 | — |
case-13 | fail→fail | 9,430 | 5,119 | -46% | 1 | 1 | 0% | 1,557 | 2,030 | +30% | 0 | 0 | — |
case-14 | fail→pass | 9,412 | 3,075 | -67% | 1 | 1 | 0% | 1,338 | 1,462 | +9% | 0 | 0 | — |
case-15 | fail→pass | 9,936 | 3,099 | -69% | 1 | 1 | 0% | 1,417 | 1,689 | +19% | 0 | 0 | — |
case-16 | fail→pass | 12,980 | 2,760 | -79% | 1 | 1 | 0% | 2,260 | 1,559 | -31% | 0 | 0 | — |
case-17 | fail→pass | 13,725 | 3,615 | -74% | 1 | 1 | 0% | 2,271 | 1,900 | -16% | 0 | 0 | — |
case-18 | pass→pass | 6,930 | 3,486 | -50% | 1 | 1 | 0% | 1,268 | 1,867 | +47% | 0 | 0 | — |
case-19 | fail→pass | 14,180 | 4,291 | -70% | 1 | 1 | 0% | 2,195 | 1,900 | -13% | 0 | 0 | — |
case-20 | pass→pass | 18,904 | 2,918 | -85% | 1 | 1 | 0% | 2,866 | 1,490 | -48% | 0 | 0 | — |
case-21 | pass→pass | 14,929 | 6,202 | -58% | 1 | 1 | 0% | 2,133 | 2,104 | -1% | 0 | 0 | — |
case-22 | fail→fail | 10,610 | 8,892 | -16% | 1 | 1 | 0% | 1,679 | 2,702 | +61% | 0 | 0 | — |
case-23 | fail→fail | 10,406 | 11,180 | +7% | 1 | 1 | 0% | 1,754 | 2,808 | +60% | 0 | 0 | — |
case-24 | fail→fail | 9,846 | 11,154 | +13% | 1 | 1 | 0% | 1,905 | 3,179 | +67% | 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. 24 cases were attempted. The headline lift of +58 percentage points is the difference between those two pass rates over the 24 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.