Sifthound
github.com/khsarvar/sifthound- Category
- Developer Tools
- Rank
- No. 2002Tools index
Previous survey · No. 1950 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Data, Retrieval & Knowledge
- Interfaces
- API · MCP
- GitHub
- 3 stars
- Latest release
- v0.3.0
- Date
About
Sifthound is an open-source, self-hosted web search API for AI agents that mirrors Tavily's search, extract, crawl, and map endpoints and response formats, so existing Tavily SDK or LangChain integrations work against it by changing only the base URL. It searches through a self-hosted SearXNG metasearch instance, removing the need for a search API key, and extracts clean markdown or text from web pages.
At a glance
Sifthound, now named Siftdog, is a search and page extraction API you run yourself. It is worth evaluating if you already call Tavily-style endpoints and want to control the service that fetches and ranks pages.
Good fit
Developers who can run a search service and want to test their existing client against it.
Consider something else when
Skip it if you want someone else to operate the search infrastructure or your queries cannot be sent to public search engines.
Start here
Pick a public query with a known answer. Compare search results and extracted text with the sources your application already uses. Check for missing pages before switching a real workflow.
Cost and requirements
MIT-licensed. Search requires a SearXNG instance with JSON enabled. Optional generated answers use an Anthropic API key. Budget for hosting and any model calls.
Limits to know
Self-hosting controls the API and extraction. SearXNG still sends searches upstream, and result coverage and rate limits depend on those engines.
Sources checked 2026-11-05. Source-based assessment.
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.