unsearch
por rakesh1002
Web & Frontend DevelopmentResumen
Search the web, scrape content, and conduct deep research using the UnSearch API.
Descripción
Qué hace esta skill
UnSearch provides web search across Google, Bing, and DuckDuckGo with optional full-page content scraping and URL extraction. It also supports fact verification and multi-source deep research with AI synthesis. It is an open-source alternative to Tavily and Exa, targeting AI agents, RAG pipelines, and LLM applications.
It offers a Tavily-compatible agent search endpoint and self-hosting option, with 5,000 free queries per month, making it a drop-in alternative without vendor lock-in.
Cuándo usarla
- Fetching current news or events for an AI agent that needs real-time data
- Extracting full article text from specific URLs for a RAG pipeline
- Verifying a factual claim against multiple live web sources
- Running deep multi-source research with an AI-synthesized executive summary
- Restricting search to specific domains to answer a focused question
Flujo de trabajo de ejemplo
User asks: Verify whether GPT-4 was released in March 2023
- 1
Sends the claim to the /verify/claim endpoint with depth set to 'thorough'
- 2
API searches across multiple web sources for supporting and contradicting evidence
- 3
Returns a verdict string, confidence score, and list of sources checked
- 4
Agent presents the verdict with confidence percentage and citations
Verdict: true (95% confidence). GPT-4 was released March 14, 2023, confirmed across 12 sources.
Requisitos
- UNSEARCH_API_KEY environment variable
- unsearch.dev account to obtain an API key
Instalación
clawhub install rakesh1002/unsearchEjecuta el comando de la CLI o pega el prompt en Claude Code, Cursor, Codex o el agente de programación que uses.