What it does
Turns web research into compact, source-grounded snippets that your MCP client can use.
TinySearch
Self-hosted web research for local AI agents.
TinySearch searches the web, crawls relevant pages, chunks content, reranks results, and returns a compact source-grounded prompt instead of raw webpages.
research(query)
scrape_url(url, query)
get_current_datetime()
search -> crawl -> chunk -> rerank -> grounded promptHow it works
Search -> crawl -> chunk -> rerank -> grounded prompt.
Use cases
Give local LLMs web search, reduce context waste, and build source-grounded MCP agents.
Install
The docs walk through the full setup path: Docker, the MCP endpoint, optional source mode, examples, and troubleshooting.
Fast path
Start the local research layer.
Run TinySearch with its own SearXNG instance as an MCP server over Streamable HTTP.
http://localhost:8000/mcpMCP tools
TinySearch returns a grounded prompt in the tool response. Your client model writes the final answer from that evidence.
Web research
research(query)
Discover relevant URLs, crawl the best pages, and return ranked evidence chunks.
Known URLs
scrape_url(url, query)
Inspect a known URL and extract query-relevant snippets from that page.
Fresh context
get_current_datetime()
Orient time-sensitive research before calling the web research tools.
Pipeline control
Configuration
Use SearXNG by default, configure DuckDuckGo fallback, embeddings, chunking, and quotas.
Example output
QUESTION
What should I know about the project?
RESULTS
1. TinySearch
https://github.com/MarcellM01/TinySearch
Self-hosted web research for MCP agents...
INSTRUCTIONS
Answer only from the results. Cite source URLs.