search

The Search skill provides an efficient interface to the Tavily web search API, specifically engineered for AI agents and LLMs. By delivering clean, relevant, and context-aware results, it enables users to retrieve real-time information, research technical topics, or stay updated on current events without needing to write custom integration code. It is an essential tool for developers, researchers, and AI users who require high-quality web-sourced data with metadata, content snippets, and relevance scores.

• Powered by Tavily's LLM-optimized engine, returning prioritized content chunks and NLP summaries.

• Supports advanced search parameters including search depth (ultra-fast to advanced), time-range filtering, and domain-specific inclusion/exclusion.

• Enables automated data retrieval for RAG (Retrieval-Augmented Generation) pipelines by providing raw content toggles and image-related metadata.

• Supports flexible authentication via either the Tavily MCP server (OAuth) or a direct API key configuration in settings.

• Ideal for use cases like market research, trend analysis, fact-checking, and gathering domain-specific technical documentation from sites like GitHub or ArXiv.

• The skill accepts structured JSON payloads containing queries, search depth, and filtering constraints.

• Users should keep search queries under 400 characters and utilize sub-queries for complex topics to ensure optimal relevance.

• The search depth parameter (basic, advanced, fast, ultra-fast) can be tuned to balance latency against result precision.

• Authentication is handled seamlessly by the MCP server; users only need to ensure their Tavily account is active.

• Constraints include a maximum of 20 results per request and limitations on domain inclusion (300) and exclusion (150) lists.