websearch — use free-search, never curl a search engine

Search engines block plain curl/httpx clients (DDG "anomaly 202", Bing CAPTCHAs, Google 429). Do not retry them with different UAs — that is the loop many sessions get stuck in. The local free-search CLI solves it: multi-engine with reciprocal-rank fusion, Chrome-fingerprinted HTTP, and automatic searx rescue when an engine walls off.

Commands

free-search search "query" [n] [--json]      # ranked link list (default n=8)
free-search fetch <url>                      # page text as markdown
free-search research "question" [--depth N]  # search+read in one call, cited brief

• --json on search → {"query", "engines", "cached", "results": [{title, url, snippet, score}]}.
• Fetches are cached locally; fetch output may note _fetched via cache_.
• Typical latency: ~5–15s per search. No API key, no config needed.

Routing