spb/tools-spboucher-ai

1{2  "slug": "tavily",3  "name": "Tavily",4  "aliases": ["Tavily Search API"],5  "vendor": { "name": "Tavily (Alpha AI Technologies Inc.)", "slug": "tavily", "website": "https://www.tavily.com/" },6  "homepage_url": "https://www.tavily.com/",7  "docs_url": "https://docs.tavily.com/",8  "repo_url": "https://github.com/tavily-ai/tavily-python",9  "pricing_url": "https://www.tavily.com/pricing",10  "product_type": "api-service",11  "product_status": "active",12  "summary": "Web search, extraction, crawling and research API designed for LLM agents, returning concise source content with SDKs, framework integrations and a hosted MCP server",13  "description": "Tavily is a web-access API built for AI agents rather than for human search pages. Its Search endpoint returns ranked results with extracted content snippets sized for a model's context; Extract pulls clean text from one or more URLs; Crawl and Map walk a site and expose its structure; and a Research endpoint produces a multi-source, cited report. Results are returned as JSON ready to drop into a prompt, which is why Tavily appears as a default search tool in LangChain, CrewAI, the Vercel AI SDK, n8n and similar frameworks.\n\nAccess is through a REST API, official Python (`tavily-python`, MIT) and JavaScript SDKs, a CLI, and a hosted MCP server that plugs into Claude, Codex, Cursor and other MCP clients. Usage is metered in credits that differ by request type; a free tier with monthly credits is offered without a card, alongside pay-as-you-go, monthly subscriptions sized by credit volume, and enterprise agreements, with programs for students and startups.\n\nTavily suits developers who need a reliable \"search the web\" tool for an agent without building crawling or ranking themselves, and researchers prototyping retrieval pipelines. It is a proprietary hosted service: the index and ranking are opaque, there is no self-hosted option, and content depends on what the service can fetch, so paywalled or login-protected pages are out of reach.",14  "capabilities": ["Agent-oriented web search with extracted content", "Clean text extraction from arbitrary URLs", "Site crawl and map endpoints", "Cited multi-source research reports", "Python and JavaScript SDKs plus CLI", "Hosted MCP server for agent clients", "Framework integrations (LangChain, CrewAI, Vercel AI SDK, n8n)"],15  "workflows": ["Give a LangGraph or CrewAI agent a search tool", "Ground a chatbot's answers in current web sources", "Fetch and clean a list of article URLs for a corpus", "Generate a sourced briefing on a topic"],16  "advantages": ["Output shaped for prompts, reducing post-processing", "Free tier and quick setup", "Wide framework and MCP support"],17  "limitations": ["Proprietary hosted service; no self-hosting", "Opaque ranking and coverage", "Cannot reach paywalled or authenticated content", "Credit costs differ by request type"],18  "prerequisites": ["API key (free tier available)"],19  "audiences": ["Agent and RAG developers", "Researchers prototyping retrieval", "Teams adding live web grounding to assistants"],20  "alternatives": ["exa", "brave-search-api", "serpapi", "firecrawl"],21  "comparison_notes": "Exa is the closest peer, also agent-first, but is built on its own neural index with semantic and similarity search and asynchronous research agents; Tavily emphasizes ready-to-use snippets and breadth of framework integrations. Brave Search API exposes an independent web index with classic result types and an LLM-context endpoint. SerpApi returns structured results scraped from Google and other engines, aimed at SEO and data collection as much as agents. Firecrawl overlaps on extraction and search but centres on crawling whole sites into Markdown.",22  "categories": [{ "slug": "search-apis", "primary": true }, { "slug": "data-collection-crawling" }],23  "tags": ["api", "llm", "agents", "rag", "web-search", "python", "typescript", "mcp"],24  "use_cases": ["search-the-web-programmatically", "build-agents", "crawl-websites"],25  "operating_systems": ["web"],26  "execution_modes": ["cloud"],27  "interfaces": ["api", "sdk", "cli", "mcp-server", "web"],28  "deployment_models": ["saas"],29  "integrations": ["LangChain", "CrewAI", "Vercel AI SDK", "n8n", "OpenAI Agents", "Mastra", "Model Context Protocol", "Claude", "Cursor"],30  "data_formats": ["JSON", "Markdown"],31  "languages": ["Python", "JavaScript", "TypeScript"],32  "source_model": "proprietary",33  "pricing_model": ["freemium", "usage-based", "paid", "enterprise", "academic"],34  "pricing_checked_at": "2026-11-02",35  "self_hostable": false,36  "sources": [37    { "url": "https://www.tavily.com/", "type": "official_site", "title": "Tavily", "supports": ["name", "summary"], "checked_at": "2026-11-02" },38    { "url": "https://docs.tavily.com/documentation/about", "type": "official_docs", "title": "Tavily documentation – About", "supports": ["capabilities", "integrations", "description"], "checked_at": "2026-11-02" },39    { "url": "https://www.tavily.com/pricing", "type": "pricing_page", "title": "Tavily pricing", "supports": ["pricing_model"], "checked_at": "2026-11-02" },40    { "url": "https://github.com/tavily-ai/tavily-python", "type": "repository", "title": "tavily-ai/tavily-python", "supports": ["languages", "vendor"], "checked_at": "2026-11-02", "note": "SDK is MIT-licensed; the service itself is proprietary" }41  ],42  "confidence": "high",43  "discovery_method": "editorial-research",44  "depth": "deep",45  "editorial_status": "reviewed",46  "visibility": "public",47  "featured": false,48  "author": "batch-C"49}50