web
Search the web via DuckDuckGo, crawl pages for content, and fetch news articles. Save results to disk using harvest to avoid high context costs.
Instructions
Web search and page fetching. Returns {success, results/content, url}. Use ONLY after exhausting local sources: retrieve() -> kg(neighbors) -> web. Actions: 'search' (DuckDuckGo, returns title/url/snippet), 'crawl' (fetch page content). Set news=true for DDG news endpoint (adds date/source fields). High context cost - prefer harvest=True to save content to disk instead of loading into context. Results cached 20min. For bot-protected sites, use fx tool instead.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| url | No | URL to crawl. Required for action='crawl'. | |
| news | No | Use DDG news endpoint instead of web. Returns date/source per result. search action only. | |
| query | No | Search query. Required for action='search'. | |
| action | Yes | 'search' for DuckDuckGo query, 'crawl' to fetch a specific URL | |
| harvest | No | Save content to disk, return metadata only. Saves context tokens. | |
| summary | No | Return summary only (default: true). Set false for full page content. | |
| timelimit | No | Time filter for news: 'd' (day), 'w' (week), 'm' (month). | |
| harvest_dest | No | Subfolder under harvested/ (e.g., 'docs/'). Required if harvest=true. | |
| timeout_seconds | No | Request timeout. Default: 30. |