Evidence-first web and workspace search with bounded results, source labels, and machine-readable provenance. Use for research questions, current facts, or finding content in local memory and skills.

Install

openclaw skills install @iyeque/iyeque-unified-web-search

Evidence-first unified search

Use index.js when a task needs web retrieval plus a bounded search of local memory/ and skills/ content. Results are JSON on stdout; warnings are JSON on stderr so callers can preserve provenance.

Quick start

bash
node index.js --query "latest climate policy" --sources '["tavily"]' --allow-network --max_results 5
node index.js --query "meeting notes" --sources '["local"]' --allow-local --max_results 10
node index.js --query "AI developments" --sources '["tavily","local"]' --max_results 10

Trust and scope

  • tavily requires TAVILY_API_KEY and sends the query to Tavily.
  • local searches only workspace memory/ and skills/, with bounded depth and file size.
  • Symlinked files and directories are skipped; binary/unreadable files are ignored.
  • web-search-plus and browser are accepted for compatibility but report that they are not configured rather than pretending to search them.
  • Queries are trimmed and capped at 500 characters; max_results is 1–50.

Each result includes a source label, title, URL when available, score, and content/snippet. Treat network results as external evidence and verify important claims before relying on them.

Source and support

Source: https://github.com/iyeque/openclaw-skills/tree/main/skills/unified-web-search

Report bugs and security concerns through the repository issue guidance.