tavily_skill

Overview

This skill lets Alice perform web searches using the Tavily API to fetch timely news, technical research, and deep reference material. It supports multiple search modes (basic, research, news) and domain inclusion/exclusion to focus results. Use it when you need current internet data or structured research outputs.

How this skill works

The skill calls the Tavily API with the provided query and selected mode. Basic mode returns concise, human-readable summaries; research mode returns richer JSON with expanded results and metadata; news mode filters results by recency (days). You can restrict or exclude specific domains and control result count with simple command-line flags.

When to use it

• Get a quick, readable summary of web results for conversational answers.
• Gather in-depth, structured data for technical research or reports.
• Fetch recent news within a specific time window for situational awareness.
• Limit searches to or exclude particular domains for curated sourcing.
• Increase result count when compiling literature or evidence.

Best practices

• Default to basic mode for ordinary Q&A to keep responses concise and fast.
• Use research mode only for complex topics or when you need detailed JSON for downstream processing.
• Set include-domains or exclude-domains to control source quality and avoid noisy sites.
• Keep max-results between 1 and 20 to balance completeness and performance.
• Store TAVILY_API_KEY in a project .env file and install the client with pip install tavily-python.

Example use cases

• Answering a user question with fresh web evidence using basic mode.
• Producing a research digest with citations and metadata by running research mode and exporting the JSON.
• Monitoring recent developments on a topic using news mode and a days window.
• Compiling authoritative sources by limiting searches to specific domains like nature.com or science.org.
• Excluding low-quality domains when assembling a literature list for a report.