About this skill

Problem

AI agents often receive noisy, weakly structured web search results when answering factual, news, competitor, or documentation questions. Tavily Search reframes search as a model-friendly interface: it returns cleaner, more relevant content that agents can extract, reason over, and cite.

How It Works

Query search: supports general and news topics; news queries can be scoped to the last n days with --days.
Result control: defaults to 5 results; -n adjusts the count, up to 20, balancing speed against coverage.
Deep retrieval: --deep uses a slower, more comprehensive advanced search for deeper research.
URL extraction: pulls readable content from a specific page, reducing HTML noise.

Limits And Notes

This capability requires a valid TAVILY_API_KEY. It is best used as a search and web-content input source for agents, not as a local code runner, database query engine, or private knowledge base. For time-sensitive news, set --topic news and --days explicitly.

Screenshots

Use Cases

When a support agent needs the latest outage notices, search the news topic with a day limit.
When comparing competitor API changes, increase result count and use deep search for broader coverage.
Before drafting an industry weekly, select key articles from results and extract URL content for summaries.
When answering time-sensitive policy questions, scope search to recent days with news topic and --days.

Best For

Engineers building support agents who need fresh notices as model context.
Analysts writing competitor research who need topic, date, and depth controls.
Editors producing news briefs who need article discovery and content extraction.
Developers integrating web search who need structured results for agents.