Awesome-omni-skill library-doc

Index and search library documentation locally for offline use. Invoke when user asks to index docs, search library topics, or list indexed libraries.

install

source · Clone the upstream repo

git clone https://github.com/diegosouzapw/awesome-omni-skill

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/diegosouzapw/awesome-omni-skill "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/development/library-doc" ~/.claude/skills/diegosouzapw-awesome-omni-skill-library-doc && rm -rf "$T"

manifest: skills/development/library-doc/SKILL.md

source content

Library Documentation Indexer

Commands

# List
python3 ./scripts/list.py                         # all libraries
python3 ./scripts/list.py <library>               # docs in a library
python3 ./scripts/list.py --delete <library>      # delete a library

# Search
python3 ./scripts/search.py "<query>"                       # all libraries
python3 ./scripts/search.py "<query>" --library <library>   # specific library
python3 ./scripts/search.py "<query>" --limit <n>           # limit results

# Read
python3 ./scripts/read.py <library> "<title>"               # by title (exact)
python3 ./scripts/read.py <library> <id>                    # by document ID
python3 ./scripts/read.py <library> "<title>" --lines 1-100 # specific lines

# All commands support --json

Search → Read Workflow

# 1. Search
python3 ./scripts/search.py "<query>" --library <library>

# 2. Read (use title or ID from search results)
python3 ./scripts/read.py <library> "<title>"
python3 ./scripts/read.py <library> <id>

# 3. For long docs, read in chunks
python3 ./scripts/read.py <library> "<title>" --lines 1-100
python3 ./scripts/read.py <library> "<title>" --lines 100-200

Indexing a Library

Check for preset first

python3 ./scripts/presets.py list              # List all available presets
python3 ./scripts/presets.py show <library>    # Show preset details

If a preset exists, use it directly (see "Using Presets" below).

Manual indexing (no preset)

Step 1: Clarify what to index

If ambiguous, ask the user:

Library/framework (ex: React, FastAPI)
Web API or spec (ex: WebSocket API)
Different implementations across languages

Step 2: Find docs source

Search:

"<library> documentation github"

"<library> docs site:github.com"

Look for:

Official repo with
```
docs/
```
,
```
content/
```
, or
```
pages/
```
folder
Dedicated docs repo (
```
<lib>-docs
```
,
```
<lib>.dev
```
)
README-only projects

For specific versions: check releases/tags (

v15.0.0

15.x

docs-v15

)

Step 3: Identify structure

Docs path:
```
docs/
```
,
```
content/
```
,
```
src/content/
```
, etc.
File types:
```
.md
```
,
```
.mdx
```
,
```
.rst
```
Base URL: live docs URL (e.g.,
```
https://<library>.dev/docs
```
)

Step 4: Clone and index

Naming:

<library>

for latest,

<library>-<version>

for specific version.

CRITICAL: Always use sparse checkout to only download docs.

cd /tmp && rm -rf <lib>-docs

# clone with sparse checkout
git clone --depth 1 --filter=blob:none --sparse <repo-url> <lib>-docs
cd <lib>-docs && git sparse-checkout set <docs-path>

# verify docs exist
find <docs-path> -type f \( -name "*.md" -o -name "*.mdx" \) | wc -l

# index
find <docs-path> -type f \( -name "*.md" -o -name "*.mdx" \) | \
  python3 ./scripts/index.py <lib> --batch --base-url "<base-url>" --strip-prefix "<docs-path>/"

# cleanup and verify
rm -rf /tmp/<lib>-docs
python3 ./scripts/list.py <lib>

Common Patterns

Standard docs folder

cd /tmp && rm -rf <lib>-docs
git clone --depth 1 --filter=blob:none --sparse <repo-url> <lib>-docs
cd <lib>-docs && git sparse-checkout set docs
find docs -type f \( -name "*.md" -o -name "*.mdx" \) | \
  python3 ./scripts/index.py <lib> --batch --base-url "<base-url>" --strip-prefix "docs/"
rm -rf /tmp/<lib>-docs

README-only project

cd /tmp && rm -rf <lib>-repo
git clone --depth 1 <repo-url> <lib>-repo
cat <lib>-repo/README.md | python3 ./scripts/index.py <lib> -f "README.md" -u "<repo-url>#readme"
rm -rf /tmp/<lib>-repo

Specific version

cd /tmp && rm -rf <lib>-docs
git clone --branch <tag> --depth 1 --filter=blob:none --sparse <repo-url> <lib>-docs
cd <lib>-docs && git sparse-checkout set <docs-path>
find <docs-path> -type f \( -name "*.md" -o -name "*.mdx" \) | \
  python3 ./scripts/index.py <lib>-<version> --batch --base-url "<base-url>" --strip-prefix "<docs-path>/"
rm -rf /tmp/<lib>-docs

Using Presets

For popular libraries, use the preset for correct repo/path/URL:

python3 ./scripts/presets.py show react   # Show preset config

# Then use the preset values:
cd /tmp && rm -rf react-docs
git clone --depth 1 --filter=blob:none --sparse https://github.com/reactjs/react.dev react-docs
cd react-docs && git sparse-checkout set src/content
find src/content -type f \( -name "*.md" -o -name "*.mdx" \) | \
  python3 ./scripts/index.py react --batch --base-url "https://react.dev" --strip-prefix "src/content/"
rm -rf /tmp/react-docs

Available presets: react, nextjs, vue, svelte, nuxt, angular, astro, remix, tailwind, shadcn, express, fastapi, django, flask, hono, nestjs, rails, laravel, bun, node, deno, typescript, python, rust, go, prisma, drizzle, supabase, redux, zustand, tanstack-query, vitest, playwright, jest, pytest, vite, langchain, openai, anthropic, docker, kubernetes, and more.

Troubleshooting

0 docs indexed: check docs path exists after sparse checkout
No .md files: try
```
.mdx
```
,
```
.rst
```
, or check actual extension
Wrong URLs: check live docs site structure
Re-indexing: just run again, documents are upserted by path
Large doc chunking: files >8KB auto-split by ## headings; use --no-chunk to disable