curated-search
by qsmtco
Web & Frontend DevelopmentBrief
Domain-restricted full-text search over curated technical documentation.
Description
What This Skill Does
Domain-restricted full-text search over a curated whitelist of technical documentation sources such as MDN and the Python docs. After an initial crawl builds a local index, all searches run offline with no external network calls. Results include title, URL, snippet, and a BM25 relevance score.
Queries run fully offline against a pre-approved domain whitelist, so results come only from authoritative sources rather than general web search noise.
When to Use It
- Looking up CSS property syntax without SEO-spam results
- Searching Python docs offline during travel
- Filtering documentation queries to a single trusted domain
- Paginating through large result sets inside an agent workflow
- Checking async/await behavior across MDN and Python references in one query
Example Workflow
User asks: How does async/await work in Python?
- 1
Invoke curated-search.search with query 'async await' and domain 'docs.python.org'
- 2
Retrieve top results ranked by BM25 relevance score
- 3
Extract title, URL, and snippet from each result
- 4
Return the highest-scoring excerpts to the user
A ranked list of excerpts from docs.python.org covering async/await syntax and usage
Installation
clawhub install qsmtco/curated-searchRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.