Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Fetch up-to-date documentation from Context7 for a library. Use when you need current API syntax, code examples, or official documentation for any library or framework.
.claude/skills/valtterimelkko-context7-docs/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 243% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -45% | 0% |
Fetch real-time, version-specific documentation and code examples for libraries.
Use this skill when:
CONTEXT7_API_KEY in the environment or a shell startup file such as ~/.bashrc (the bundled scripts can read either)owner/repo format (use context7-search first if unknown)bashpython3 ./scripts/fetch_docs.py --library "owner/repo"
bashpython3 ./scripts/fetch_docs.py --library "owner/repo" --topic "TOPIC"
bashpython3 ./scripts/fetch_docs.py --library "owner/repo" --type info
bashpython3 ./scripts/fetch_docs.py --library "owner/repo" --version "v15.1.8"
| Parameter | Required | Default | Description | |-----------|----------|---------|-------------| | --library | Yes | - | Library in owner/repo format | | --topic | No | - | Focus on specific feature/topic | | --type | No | code | code for snippets, info for docs | | --version | No | latest | Specific version tag | | --page | No | 1 | Page number for pagination |
bashpython3 ./scripts/fetch_docs.py \ --library "vercel/next.js" \ --topic "routing"
bashpython3 ./scripts/fetch_docs.py \ --library "prisma/prisma" \ --topic "postgresql connection"
bashpython3 ./scripts/fetch_docs.py \ --library "facebook/react" \ --topic "hooks" \ --type info
bashpython3 ./scripts/fetch_docs.py \ --library "vercel/next.js" \ --version "v14.3.0" \ --topic "app router"
json{ "success": true, "data": { "library": "vercel/next.js", "topic": "routing", "type": "code", "content": { "content": "### Dynamic Route Handler\n\nSource: ...\n\n```typescript\nexport async function GET(...) {...}\n```\n\n---\n\n### Another Example\n..." } } }
| Library | ID | Common Topics | |---------|-------------|---------------| | Next.js | vercel/next.js | routing, server components, api routes | | React | facebook/react | hooks, state, context, suspense | | Prisma | prisma/prisma | setup, queries, relations, migrations | | Supabase | supabase/supabase | auth, realtime, row level security | | FastAPI | fastapi/fastapi | routing, validation, dependencies | | Express | expressjs/express | middleware, routing, error handling | | Tailwind | tailwindlabs/tailwindcss | configuration, utilities, plugins |
User: "How do I set up authentication with Supabase?"
1. Search for library (if ID unknown):
python3 ./scripts/resolve_library.py --query "supabase"
→ Returns supabase/supabase
2. Fetch focused documentation:
python3 ./scripts/fetch_docs.py \
--library "supabase/supabase" \
--topic "authentication"User: "Show me the latest Next.js routing API"
Fetch directly (skip search):
python3 ./scripts/fetch_docs.py \
--library "vercel/next.js" \
--topic "app router"Library not found (404):
context7-search to find the correct library IDowner/repo (e.g., vercel/next.js)Topic not matched:
Authentication error:
--topic--type code gives practical examples--type info for documentation/explanations--version for version-specific syntax| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,808 | 11,526 | -27% | 1 | 1 | 0% | 3,750 | 1,541 | -59% | 0 | 0 | — |
case-02 | fail→fail | 17,655 | 3,886 | -78% | 1 | 1 | 0% | 2,903 | 1,442 | -50% | 0 | 0 | — |
case-03 | fail→fail | 7,591 | 7,536 | -1% | 1 | 1 | 0% | 1,871 | 1,612 | -14% | 0 | 0 | — |
case-04 | fail→fail | 12,254 | 7,562 | -38% | 1 | 1 | 0% | 2,646 | 1,476 | -44% | 0 | 0 | — |
case-05 | fail→pass | 5,420 | 2,052 | -62% | 1 | 1 | 0% | 452 | 1,550 | +243% | 0 | 0 | — |
case-06 | fail→fail | 14,447 | 3,779 | -74% | 1 | 1 | 0% | 2,312 | 1,435 | -38% | 0 | 0 | — |
case-07 | fail→pass | 15,198 | 3,950 | -74% | 1 | 1 | 0% | 3,105 | 1,453 | -53% | 0 | 0 | — |
case-08 | fail→pass | 11,436 | 3,526 | -69% | 1 | 1 | 0% | 2,663 | 1,412 | -47% | 0 | 0 | — |
case-09 | fail→pass | 8,926 | 1,490 | -83% | 1 | 1 | 0% | 1,574 | 1,413 | -10% | 0 | 0 | — |
case-10 | fail→fail | 11,966 | 4,576 | -62% | 1 | 1 | 0% | 2,500 | 1,482 | -41% | 0 | 0 | — |
case-11 | fail→fail | 14,981 | 6,037 | -60% | 1 | 1 | 0% | 2,638 | 1,640 | -38% | 0 | 0 | — |
case-12 | fail→pass | 14,707 | 1,489 | -90% | 1 | 1 | 0% | 2,590 | 1,420 | -45% | 0 | 0 | — |
case-13 | fail→fail | 13,763 | 4,524 | -67% | 1 | 1 | 0% | 3,008 | 1,377 | -54% | 0 | 0 | — |
case-14 | fail→fail | 21,477 | 5,136 | -76% | 1 | 1 | 0% | 2,937 | 1,452 | -51% | 0 | 0 | — |
case-15 | fail→fail | 12,920 | 8,461 | -35% | 1 | 1 | 0% | 2,547 | 2,151 | -16% | 0 | 0 | — |
case-16 | fail→fail | 11,619 | 6,044 | -48% | 1 | 1 | 0% | 2,712 | 1,466 | -46% | 0 | 0 | — |
case-17 | fail→fail | 17,085 | 1,418 | -92% | 1 | 1 | 0% | 3,198 | 1,426 | -55% | 0 | 0 | — |
case-18 | fail→pass | 13,869 | 2,400 | -83% | 1 | 1 | 0% | 2,712 | 1,478 | -46% | 0 | 0 | — |
case-19 | fail→pass | 7,912 | 2,008 | -75% | 1 | 1 | 0% | 1,529 | 1,597 | +4% | 0 | 0 | — |
case-20 | pass→pass | 7,814 | 3,022 | -61% | 1 | 1 | 0% | 1,453 | 1,786 | +23% | 0 | 0 | — |
case-21 | fail→fail | 3,033 | 5,258 | +73% | 1 | 1 | 0% | 470 | 1,404 | +199% | 0 | 0 | — |
case-22 | pass→pass | 8,908 | 7,010 | -21% | 1 | 1 | 0% | 1,639 | 2,487 | +52% | 0 | 0 | — |
case-23 | pass→pass | 5,826 | 4,553 | -22% | 1 | 1 | 0% | 1,038 | 2,036 | +96% | 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 11 counted toward the lift figure. The other 12 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 +30 percentage points is the difference between those two pass rates over the 11 comparable cases. 1 case got worse with the skill loaded, and it is 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.