From 212708b464c2bdbdd9670c00804623dca0729d2c Mon Sep 17 00:00:00 2001 From: Agando Date: Wed, 27 May 2026 06:00:43 +0200 Subject: auto: 2026-05-27_06:00 changes --- skills/web-search-and-fetch.md | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+) create mode 100644 skills/web-search-and-fetch.md (limited to 'skills/web-search-and-fetch.md') diff --git a/skills/web-search-and-fetch.md b/skills/web-search-and-fetch.md new file mode 100644 index 0000000..fcd5f91 --- /dev/null +++ b/skills/web-search-and-fetch.md @@ -0,0 +1,27 @@ +--- +name: web-search-and-fetch +description: How to search the web and fetch URL content using the WebX11 proxy API +tags: [search, fetch, web, proxy, api] +category: research +--- +## Steps +1. **Search the web:** + - Call web_search(query) — uses /api/search which queries SearXNG + DDG fallback + - Results include: title, url, content snippet, source engine + - Always available (same-origin, no CORS issues) + - 5-minute server-side cache for repeated queries + +2. **Fetch URL content:** + - Call web_fetch(url) — uses /api/extract which strips HTML, extracts clean text + - Returns: title, content (up to 4000 chars), word_count, extraction method + - SSRF protection blocks internal network addresses + - 15-minute server-side cache + +3. **Search flags (via terminal search command):** + - search --summarize — adds AI-generated overview via Ollama + - search --fetch — fetches content snippets for top 3 results + +4. **When to use:** + - User asks about current events, facts, research → call web_search + - User says "read this article" or provides a URL → call web_fetch + - For in-depth analysis: search first, then fetch top results \ No newline at end of file -- cgit v1.2.3