Searches the web, Wikipedia, and scholarly indexes and extracts readable page text.
npx skills add https://github.com/BTreeMap/SKILLs --skill search-web-btreemap
Please help me install this Agent Skill. Skill: search-web Source: https://github.com/BTreeMap/SKILLs/tree/main/search-web Command: npx skills add https://github.com/BTreeMap/SKILLs --skill search-web-btreemap
web for open questions, instant for definitions, wiki for encyclopedic summaries, scholar for papers with DOI, year, and citation counts, and fetch for the readable text of one page. - Scholarly source selection: Choose OpenAlex for cross-field coverage, Crossref for DOI registry lookups, or arXiv for preprints with fielded queries. - Caching and safety: Repeated queries are answered from a local cache, fetched content is treated as untrusted data, and prompt-injection text found in pages is reported rather than followed. - Use Case: Ask your agent to find recent papers on a topic by DOI, then fetch the full text of the most relevant result to verify a claim before citing it. ## Quick Start Ask the agent to search the scholarly record for papers on your topic and fetch the full text of the top result.High-intent search queries and answers about installing and using this skill.
Use the web channel, which queries multiple search engines through the ddgs library with automatic fallback between backends. Results return as JSON with title, URL, snippet, and source fields, and no API key is required.
Run the scholar channel with --source set to openalex, crossref, or arxiv. OpenAlex spans every field and is the default, Crossref is the DOI registry, and arXiv covers preprints and ranks fielded queries like all:"exact phrase" better than bare terms.
Yes, the fetch channel extracts the main article text from a page using trafilatura. It refuses PDFs, paywalled pages, listings, and JavaScript-rendered pages rather than returning empty or mangled output.
No, fetch explicitly refuses PDF URLs and directs you to the read-pdf skill instead. Use a dedicated PDF extraction skill to pull text and metadata from PDF files with page-cited evidence.
The web channel uses unofficial, rate-limited backends, so empty results can occur under throttling. The skill suggests retrying later or switching to the wiki or scholar channels, which use official keyless APIs.
Wikipedia rows are tertiary summaries suitable for orientation but never acceptable as citations. Each wiki result includes the page summary, and claims should be verified against primary sources found via scholar or fetch.
wsxwj123Evaluate scientific claims and evidence quality using structured critique criteria.
MaraziiSynthesize verified sources into structured, fact-checked literature reviews.
shenxingyExtract numeric claims from blog posts and verify them against cited URLs.
jonhill90Fetch current library documentation from MCP sources.