tavily-tools

The tavily-tools skill provides a robust interface for interacting with Tavily AI services directly from your development environment. It is designed for engineers, researchers, and data analysts who require precise, context-aware web data retrieval and processing. By abstracting the complexity of API interactions into a unified Python CLI, it enables seamless integration of web-based intelligence into automated workflows. Whether you are performing a quick lookup for library documentation, crawling a complex technical site structure, or orchestrating a comprehensive deep-research multi-step investigation, this tool provides the necessary control over search depth, result filtering, and output formatting.

• Advanced web searching with support for domain whitelists, blacklists, and localized country filters.

• URL content extraction providing clean markdown or raw text outputs, ideal for feeding into LLM context windows.

• Website crawling and link structure mapping with configurable depth, breadth, and page limits to ensure efficient data gathering.

• Deep research agent capability that executes multi-step operations to generate structured, source-cited investigative reports.

• Unified environment configuration through TAVILY_API_KEY, supporting secure 1Password injection for seamless secret management.

• Requires python and uv for environment management and script execution.

• Commands like search, extract, crawl, map, and research are invoked with specific flags (e.g., --query, --urls, --url) to enforce correct API usage.

• Designed for high-reliability scenarios; commands provide structured JSON responses containing success status, raw results, or error diagnostics.

• Users should monitor rate limits and respect site crawling policies when using the crawl and map features to avoid excessive traffic.

• Output formats can be customized via CLI flags (e.g., --format markdown) to suit downstream processing requirements in technical pipelines.