| name | tavily |
| description | Perform AI-powered web searches to find documentation, troubleshoot errors, and summarize information. |
Tavily Web Search
This skill provides an AI-optimized web search engine via Tavily. It excels at finding recent information, exact error solutions, and synthesizing content from multiple sources.
Requires TAVILY_API_KEY to be set in your environment.
Available Tools
tavily_search(query, ...)
Search the web and retrieve ranked results with titles, URLs, and relevant content snippets.
- When to use:
- You encounter an error message you don't recognize.
- You need to look up documentation that isn't available in Context7 or your local files.
- You need to check for recent news or library updates.
- Key Parameters:
query: The search string (e.g., pydantic v2 migration guide).search_depth: Use 'basic' (default) for speed, or 'advanced' for a deeper crawl with richer content.topic: Use 'general' (default) or 'news' to favor recent articles.include_answer: Set to true to have Tavily's AI generate a summary answer of the results.include_raw_content: Set to true if you need the full HTML/text content of the resulting pages instead of just snippets.include_domains / exclude_domains: Filter results to specific sites (e.g., ["github.com", "stackoverflow.com"]).
Workflow
- Keep queries specific. Instead of searching
python docs, search python 3.12 asyncio task groups.
- If the snippets returned by a basic search aren't enough, you can re-run the search with
include_raw_content: true or search_depth: 'advanced'.
- For finding official documentation, consider using the
include_domains parameter to restrict results to docs.python.org or developer.mozilla.org.
