Tavily Search - Setup Guide¶

Source: tavily-ai/skills (25,700+ installs) Category: Web Search / Research Quality Tier: 🟢 Production

Official Tavily CLI for web search returning LLM-optimized results with content snippets, relevance scores, and metadata. Supports domain filtering, time ranges, and multiple search depths (ultra-fast to advanced). This is the authoritative Tavily integration - not the community OpenClaw wrapper.

Installation¶

# Install Tavily CLI
curl -fsSL https://cli.tavily.com/install.sh | bash

# Login with API key
tvly login

Prerequisites¶

Requirement	Details
Tavily API key	Sign up at tavily.com (free tier: 1,000 searches/month)
curl	For installer script
tvly CLI	Installed via the command above

Key Capabilities¶

Search Depths¶

Depth	Speed	Use Case
`ultra-fast`	<1s	Quick fact checks, definitions
`fast`	~1s	General knowledge, news
`basic` (default)	~2s	Balanced depth and speed
`advanced`	~3-5s	Deep research, comprehensive coverage

Topics¶

Topic	Best For
`general` (default)	Broad web search
`news`	Recent events, breaking stories
`finance`	Stock data, earnings, SEC filings

Quick Start¶

# Basic search
tvly search "quantum computing breakthroughs 2026" --json

# Advanced deep search with more results
tvly search "AI agent frameworks comparison" --depth advanced --max-results 10 --json

# Recent news only
tvly search "OpenAI announcements" --time-range week --topic news --json

# Domain-filtered search
tvly search "SEC filings" --include-domains sec.gov,reuters.com --json

# Include full page content in results
tvly search "react hooks tutorial" --include-raw-content --max-results 3 --json

# Exclude domains
tvly search "climate change" --exclude-domains wikipedia.org --json

Options Reference¶

Option	Description
`--depth`	`ultra-fast`, `fast`, `basic` (default), `advanced`
`--max-results`	0-20 (default: 5)
`--topic`	`general` (default), `news`, `finance`
`--time-range`	`day`, `week`, `month`, `year`
`--include-domains`	Comma-separated domain list
`--exclude-domains`	Comma-separated domain list
`--include-raw-content`	Include full page content in results
`--json`	Machine-readable JSON output

Hermes Integration¶

For CorpusIQ growth operations, the Tavily workflow follows this pipeline:

search → extract → map → crawl → research

Growth Use Cases¶

# Competitor monitoring
tvly search "CorpusIQ competitor AI business intelligence" --depth advanced --topic news --time-range month --json

# Market research
tvly search "business operators AI tools pain points" --depth advanced --max-results 10 --json

# Content research for social
tvly search "ecommerce analytics trends 2026" --time-range week --topic news --json

# Lead discovery
tvly search "shopify merchants looking for analytics" --depth advanced --json

Tips¶

Always use --json for machine-readable output (Hermes agents parse programmatically)
Start with basic depth for exploration, escalate to advanced for final research
Combine --topic news with --time-range week for trending content
Tavily results are LLM-optimized - cleaner and more relevant than raw Google scraping

Troubleshooting¶

Issue	Solution
`tvly: command not found`	Re-run the install script: `curl -fsSL https://cli.tavily.com/install.sh \\| bash`
Auth errors	Run `tvly login` and paste your API key
Rate limiting	Free tier: 1,000 searches/month. Upgrade at tavily.com
Empty results	Broaden query, reduce `--max-results`, or use `--depth advanced`

Integration with CorpusIQ¶

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects