web_fetch.md

 1Fetch web content (for sub-agents). Converts HTML to markdown.
 2
 3<usage>
 4- Provide URL to fetch
 5- Returns content as markdown
 6- Use when following links during research
 7</usage>
 8
 9<behavior>
10- Converts HTML to markdown for analysis
11- Large pages (>50KB) saved to temp file
12- Use grep/view on temp files for large content
13</behavior>
14
15<limits>
16- Max: 5MB
17- HTTP/HTTPS only
18- No auth/cookies
19</limits>
20
21<tip>
22Only fetch pages needed to answer the question. Don't fetch unnecessarily.
23</tip>