tavily-research

Generates cited research reports from web sources using the Tavily CLI.

2|1|Updated Jun 9, 2026
One-click install
npx skills add https://github.com/zester4/zilmate --skill tavily-research-zester4
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tavily-research
Source: https://github.com/zester4/zilmate/tree/main/.agents/skills/tavily-research
Command: npx skills add https://github.com/zester4/zilmate --skill tavily-research-zester4

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Deep research normally requires manually searching many sources, reading them, and synthesizing findings with citations. This Skill automates that workflow through the Tavily CLI, producing a structured, citation-backed report in 30-120 seconds. ## Core Features & Use Cases - Multi-source research reports: Runs tvly research to gather, analyze, and synthesize web sources into a cited report. - Model and output control: Choose mini, pro, or auto models, stream progress, define JSON output schemas, and pick citation formats (numbered, MLA, APA, Chicago). - Async workflows: Start long research jobs with --no-wait, then poll status and retrieve results by request ID. - Use Case: Ask for a comparison of AI agent frameworks and receive a saved Markdown report with numbered citations via tvly research "AI agent frameworks comparison" --model pro -o report.md. ## Quick Start Ask the assistant to research a topic such as "competitive landscape of AI code assistants" using the tavily-research skill and save the cited report to a file.

Frequently Asked Questions about tavily-research

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run deep research with citations using the Tavily CLI?

Run `tvly research "your topic"` to generate a cited report. Add `--model pro` for comprehensive analysis, `--stream` for real-time progress, or `-o report.md` to save the output to a file.

What is the difference between tavily research and tavily search?

Tavily search returns quick results for simple fact-finding, while tavily research performs deep multi-source synthesis with citations over 30-120 seconds. Use search for quick facts and research for comparisons, market reports, or literature reviews.

Which Tavily research model should I use: mini, pro, or auto?

Use mini for single-topic targeted questions (~30s), pro for comprehensive multi-angle analysis like X vs Y comparisons (~60-120s), and auto to let the API choose based on query complexity.

How do I run a long Tavily research job asynchronously?

Start with `tvly research "topic" --no-wait --json` to get a request_id immediately. Then check progress with `tvly research status <request_id>` and retrieve results with `tvly research poll <request_id> --json`.

What should I do if the tvly command is not found?

Install the Tavily CLI first by running `curl -fsSL https://cli.tavily.com/install.sh | bash && tvly login`. Do not skip this step or fall back to other tools, as the skill depends on the tvly binary being on PATH.