Tavily Search

Tavily Search

AI-powered web search that returns synthesized answers with citations and structured results.

When to Use

• User explicitly requests "use Tavily" or "search with Tavily"
• Research tasks requiring comprehensive, cited answers
• Fact-checking with source verification
• Alternative to Brave/Perplexity when specifically requested

Requirements

• Python 3.6+
• tavily-python package: pip install tavily-python
• TAVILY_API_KEY environment variable

Get your API key at: https://tavily.com

Script Location

The search script is located at skills/tavily-search/scripts/search.py relative to the application working directory. IMPORTANT: Always use the FULL path skills/tavily-search/scripts/search.py — do NOT shorten to scripts/search.py.

Quick Start

Basic search

python skills/tavily-search/scripts/search.py "latest AI developments"

Advanced search with more results

python skills/tavily-search/scripts/search.py "quantum computing breakthroughs" --max-results 10 --depth advanced

Text format output

python skills/tavily-search/scripts/search.py "climate change news" --format text

Usage

Command-line options

• query (required): Search query
• --max-results N: Maximum results (1-10, default: 5)
• --depth [basic|advanced]: Search depth (default: basic)
• basic: Fast, general search
• advanced: Deep, comprehensive search (slower, costs more)
• --no-answer: Exclude AI-generated answer
• --raw-content: Include full page content
• --format [json|text]: Output format (default: json)

Response format

Tavily returns:

• answer: AI-synthesized answer to the query
• results: Array of search results with:
• title: Page title
• url: Source URL
• content: Relevant excerpt
• score: Relevance score
• query: Original query
• response_time: API response time

Examples

Research query

python skills/tavily-search/scripts/search.py "What are the latest developments in quantum computing?" --depth advanced

Returns comprehensive answer with 5-10 cited sources.

Quick fact check

python skills/tavily-search/scripts/search.py "When was the Eiffel Tower built?" --max-results 3

Fast answer with minimal sources.

News search

python skills/tavily-search/scripts/search.py "breaking news today" --format text

Human-readable format with headlines and summaries.

Integration in OpenClaw

When called by the agent, run the script and parse the JSON output. IMPORTANT: The script path MUST be skills/tavily-search/scripts/search.py (full path from the application working directory).

import subprocess
import json

SEARCH_SCRIPT = "skills/tavily-search/scripts/search.py"

result = subprocess.run(
    ["python", SEARCH_SCRIPT, query, "--max-results", "5"],
    capture_output=True,
    text=True
)
response = json.loads(result.stdout)

Notes

• API key: Set TAVILY_API_KEY in environment or ~/.openclaw/.env
• Rate limits: Check Tavily pricing page for your plan's limits
• Search depth: Use basic for most queries; advanced for research
• Cost: Advanced search costs more credits per query

Troubleshooting

Module not found

pip install tavily-python

Missing API key

export TAVILY_API_KEY="tvly-xxx"

Permission denied

chmod +x skills/tavily-search/scripts/search.py

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: kyopark2014
• Source: kyopark2014/agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.