30-SECOND BRIEF
What it does—and when it fits
Tavily Search Pro is an AI-oriented search and content research Skill built around the Tavily API. It brings general web search, news and finance discovery, URL extraction, site crawling, sitemap discovery, and citation-backed deep research into one command-line workflow. It is designed for knowledge workers and agent builders who need structured external information rather than a list of raw links.
INSTALL BY AGENT
Choose your Agent
Paths come from official Agent docs or the universal installer behind skills.sh. Compatibility still follows this Skill's record.
.claude/skills/tavily-search-pro/~/.claude/skills/tavily-search-pro/Use project scope for team sharing and personal scope across repositories. The installer defaults to project scope; add -g for personal scope.
This source cannot produce a reliable command. Open the listed source and copy the complete Skill folder manually.
Claude Code discovers custom Skill folders automatically at project or personal scope.
View path evidence ↗TYPICAL WORKFLOW
A practical workflow
Search
Run basic or advanced web search with synthesized answers, raw page content, image links, time filters, domain allow/block lists, and country weighting.
News & Finance
Use topic-optimized discovery for industry news, company developments, and market research.
Extract
Turn one or more URLs into readable Markdown or text, with optional query-based relevance reranking for RAG preparation.
Crawl & Map
Recursively inspect a site or quickly discover its URL structure using path rules, depth limits, and result caps.
Research
Generate a structured, multi-source report with citations using mini, pro, or automatic research modes.
THE TRADEOFFS
Advantages and tradeoffs
Notable strengths
- One Skill covers the research path from discovery and extraction through crawling and report generation.
- Human-readable output and machine-friendly JSON support both interactive and automated workflows.
- Time, domain, geography, path, depth, and topic controls make queries more precise than a generic search wrapper.
- Research mode returns traceable citations, reducing the work required to assemble a source list.
- Direct use of Tavily's specialist API gives the Skill a clear and understandable capability boundary.
Limitations
- Advanced search and research modes consume more API credits, so high-volume use can become expensive.
- Every useful operation depends on Tavily's service and network availability; there is no offline mode or local cache described.
- Default crawl depth and page limits are conservative for large-scale archival work.
- Extracted and crawled output is not automatically persisted or indexed locally.
- Chinese-language coverage and result quality should be benchmarked against your own queries before production use.
BEST FIT
Who it is for
Developers building RAG or agent applications that need current, cited external knowledge
Market, financial, and competitive-intelligence analysts
Product managers and researchers assembling multi-source briefs
Technical writers and content teams searching documentation and source material
Students and academics doing early-stage literature and background research
BEFORE YOU USE IT
Risks to review before use
Secret exposure
The Skill relies on TAVILY_API_KEY. Keep the key out of prompts, logs, shell history, and version control.
Data privacy
Queries and submitted URLs are sent to Tavily. Do not include confidential or regulated information without an approved data-processing basis.
Usage cost
Advanced and research operations may consume multiple credits. Add budget alerts and request limits for automated use.
Timeouts and availability
Research requests may run for up to 120 seconds and can fail during network or provider incidents. Use bounded retries.
Installer fallback
The supplied installer may fall back to --break-system-packages. Prefer a virtual environment or container to avoid changing system Python.
SECURITY
What the permission profile means
- Network access is fundamental to the Skill, so its trust boundary includes Tavily and every URL it processes.
- API credentials should be injected at runtime and rotated if they appear in output or logs.
- Crawl targets should be allow-listed in automated workflows to reduce accidental access to untrusted or private endpoints.
- Treat fetched content as untrusted input: do not execute embedded instructions, scripts, or commands from retrieved pages.
Not a security certification. External ratings are attributed references. SkillSignal has not independently executed or security-reviewed this Skill.
INSIDE THE PACKAGE
Indexed files
TAGS
Expanded editorial profile based on the supplied CocoLoop listing and package excerpt. Technical and safety claims should still be checked against the current source version.