Sifthound — open-source, self-hosted Tavily alternative
Sifthound is an open-source, self-hosted web search API for AI agents and LLM apps, and a drop-in
replacement for the Tavily API. It serves the same /search,
/extract, /crawl and /map endpoints with the same request and response shapes, so code
written for Tavily (including the official Python SDK and the LangChain integration) works
against your own server by changing only the base URL. It needs no search API key.
- Search through a SearXNG metasearch instance, with no search API keys
- Extraction of clean markdown or text from web pages with trafilatura
- Ranking: BM25 relevance blended with the upstream engine's order;
advanceddepth fetches each page and returns its most relevant chunks - Answers (
include_answer) written by Claude from the retrieved results - Crawling and site maps with depth, breadth, limit and regex path/domain filters
- MCP server for Claude Code, Claude Desktop, Cursor and other MCP clients, over HTTP at
/mcpor stdio withsifthound mcp - SSRF protection: private and internal addresses are blocked, including via redirects and DNS rebinding
- MIT licensed
Use it as a drop-in Tavily replacement
Point the official Tavily clients at your Sifthound server with api_base_url. The key can be any
string when auth is disabled, or one of your API_KEYS.
from tavily import TavilyClient
client = TavilyClient(api_key="your-sifthound-key", api_base_url="http://localhost:8000")
results = client.search("latest python release", search_depth="advanced", max_results=5)
pages = client.extract(urls=["https://en.wikipedia.org/wiki/Okapi_BM25"])
LangChain, through langchain-tavily:
from langchain_tavily import TavilySearch
search = TavilySearch(
max_results=5, tavily_api_key="your-sifthound-key", api_base_url="http://localhost:8000"
)
search.invoke({"query": "what is BM25 ranking"})
Tested with tavily-python 0.8.4 (search, extract, crawl, map; sync and async) and
langchain-tavily 0.2.18 (search, extract).
Quick start
Docker Compose (includes SearXNG)
The full stack, with a SearXNG instance for /search, using the published image:
git clone https://github.com/khsarvar/sifthound && cd sifthound
cp .env.example .env # optional: set API_KEYS and ANTHROPIC_API_KEY
docker compose up
Try a search:
curl -s localhost:8000/search -H "Content-Type: application/json" \
-d '{"query": "latest python release", "search_depth": "advanced"}'
If you set API_KEYS, add -H "Authorization: Bearer <key>". OpenAPI docs are served at
http://localhost:8000/docs.
Docker image only
docker run -p 8000:8000 ghcr.io/khsarvar/sifthound
/extract, /crawl and /map work on their own. For /search, point it at a SearXNG
instance with the JSON format enabled: -e SEARXNG_URL=http://your-searxng:8080. Images are
published for linux/amd64 and linux/arm64, tagged latest and by version (0.1.0, 0.1).
pip
pip install sifthound
SEARXNG_URL=http://your-searxng:8080 sifthound --port 8000
Configuration is read from environment variables or a .env file (see
Configuration).
Use with MCP clients (Claude, Cursor, ...)
Sifthound is also an MCP server with four read-only tools:
sifthound_search, sifthound_extract, sifthound_crawl and sifthound_map.
Connect to a running Sifthound server (Streamable HTTP at /mcp). With Claude Code:
claude mcp add --transport http sifthound http://localhost:8000/mcp \
--header "Authorization: Bearer <key>" # omit the header if API_KEYS is empty
Or run it locally over stdio with uv, no server needed.
SEARXNG_URL is only needed for sifthound_search:
claude mcp add sifthound -e SEARXNG_URL=http://your-searxng:8080 -- uvx sifthound mcp
Claude Desktop (claude_desktop_config.json), Cursor (.cursor/mcp.json) and most other clients
take the same command as JSON:
{
"mcpServers": {
"sifthound": {
"command": "uvx",
"args": ["sifthound", "mcp"],
"env": { "SEARXNG_URL": "http://your-searxng:8080" }
}
}
}
API keys work as for the REST API: send Authorization: Bearer <key>, or append
?api_key=<key> to the URL for clients that can't set headers (URLs can end up in logs, so
prefer the header). The HTTP endpoint only answers requests addressed to localhost unless you
list your hostname in MCP_ALLOWED_HOSTS.
Sifthound vs Tavily, Firecrawl and crw
| Sifthound | Tavily | Firecrawl | crw | |
|---|---|---|---|---|
| License | MIT | Proprietary (hosted service) | AGPL-3.0 | AGPL-3.0 |
| Self-hosted | Yes | No | Yes | Yes (also a managed API) |
| Tavily-compatible API | Yes | — | No (own API) | No (own API) |
| Search source | SearXNG metasearch | Proprietary | — | — |
| JavaScript rendering | No (static HTML) | — | Yes | — |
| MCP server | Yes (HTTP and stdio) | Yes | Yes | Yes |
| Language | Python | — | TypeScript | Rust |
When to pick something else: if you'd rather not run infrastructure, or you want Tavily's neural reranking, use hosted Tavily. If you need JavaScript-rendered pages or a scraping platform with more features, look at Firecrawl or crw. Sifthound is for teams that want the Tavily API on their own servers under a permissive license.
FAQ
What is Sifthound?
Sifthound is an open-source web search and extraction API for AI agents. It reproduces the Tavily
API (/search, /extract, /crawl, /map) on infrastructure you run yourself, using SearXNG
for search results, trafilatura for content extraction and BM25 for relevance ranking.
Is Sifthound a drop-in replacement for Tavily?
For the four core endpoints, yes. The request and response fields match Tavily's, and the
official tavily-python SDK and langchain-tavily work by setting api_base_url. The
differences: relevance scores come from BM25 rather than a neural reranker, instructions
(crawl/map) and include_image_descriptions (search) are accepted but ignored, and Tavily's
/research endpoint isn't implemented.
Do I need a search API key?
No. Search results come from SearXNG, which queries public search engines. The only optional
key is ANTHROPIC_API_KEY, used when a request sets include_answer.
Does it work with LangChain?
Yes, through the official langchain-tavily package. Pass api_base_url pointing at your
Sifthound server, as in the example above.
Does Sifthound have an MCP server?
Yes. The API server exposes MCP over Streamable HTTP at /mcp, and uvx sifthound mcp runs it
over stdio for local clients such as Claude Desktop and Cursor. See
Use with MCP clients.
Is it safe to expose Sifthound on a public server?
Set API_KEYS so only your clients can call it. /extract and /crawl fetch caller-supplied
URLs, so Sifthound refuses private, loopback and link-local addresses, checked on every redirect and
at connect time against the exact address used, which also stops DNS rebinding.
Can I run it without Docker?
Yes: pip install sifthound, then run sifthound with SEARXNG_URL pointing at any SearXNG
instance with the JSON output format enabled. /extract, /crawl and /map work without
SearXNG.
Local development
python3 -m venv .venv && .venv/bin/pip install -e ".[dev]"
docker compose up searxng -d # uncomment its `ports:` in docker-compose.yml first
.venv/bin/sifthound # http://127.0.0.1:8000, docs at /docs
.venv/bin/pytest # offline test suite
.venv/bin/ruff check . && .venv/bin/ruff format .
API
All endpoints take JSON POST bodies and a Authorization: Bearer <key> header (a legacy
api_key body field is also accepted). If API_KEYS is empty, auth is disabled.
| Endpoint | Purpose | Key parameters |
|---|---|---|
/search |
Web search | query, search_depth (basic/advanced), topic (general/news), time_range, max_results, chunks_per_source, include_answer, include_raw_content, include_images, include_domains, exclude_domains |
/extract |
Clean content from up to 20 URLs | urls, extract_depth, format (markdown/text), include_images |
/crawl |
Crawl a site and return page content | url, max_depth, max_breadth, limit, select_paths, exclude_paths, select_domains, exclude_domains, allow_external, format |
/map |
List a site's URLs without content | same traversal parameters as /crawl |
Differences from hosted Tavily: instructions (crawl/map) and include_image_descriptions are
accepted but ignored; relevance scores come from BM25, not a neural reranker.
Configuration
Environment variables (see .env.example): API_KEYS, SEARXNG_URL, ANSWER_ENABLED,
ANSWER_MODEL, ANSWER_EFFORT, FETCH_TIMEOUT, FETCH_CONCURRENCY, FETCH_MAX_BYTES,
ALLOW_PRIVATE_NETWORKS, CRAWL_MAX_LIMIT, MCP_ALLOWED_HOSTS.
MCP_ALLOWED_HOSTS lists the hostnames the /mcp endpoint answers besides localhost, for
example search.example.com,search.example.com:*. Requests addressed to any other host get
421, which protects a local server from DNS-rebinding attacks by web pages.
Security: /extract and /crawl make the server fetch caller-supplied URLs. Requests to
private, loopback and link-local addresses are blocked (including via redirects) unless
ALLOW_PRIVATE_NETWORKS=true. The check runs at connect time against the exact address being
connected to, so DNS rebinding can't get around it. Fetches of user URLs ignore
HTTP(S)_PROXY, since a proxy would hide the destination address. As defense in depth for
hostile multi-tenant deployments, also restrict egress at the network level.
License
Sifthound is released under the MIT License. The "Sifthound" name is covered separately by the trademark policy: use the code freely, but forks and hosted services need a different name.
Release files for sifthound 0.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| sifthound-0.2.0.tar.gz | 27.3 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| sifthound-0.2.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 55.1 kB
Release files / sifthound-0.2.0.tar.gz
| Download URL | sifthound-0.2.0.tar.gz |
|---|---|
| Size | 27.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
fdfe9fce601c7c2fc7e173074ed693ca35de95f8c5596d6e769ba9f1bfde2fc1
|
|
BLAKE2b-256 checksum How to use checksums |
fe58666da609bc94dbef350ba05b785cad0f1a3b7e6a578c1cc6abb332ae4f94
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 25, 2026.
Transparency logRelease files / sifthound-0.2.0-py3-none-any.whl
| Download URL | sifthound-0.2.0-py3-none-any.whl |
|---|---|
| Size | 27.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
082ff54261472b0ad14d5930458d1cbc0c0899cc9fb16df7f4b0134c6acff526
|
|
BLAKE2b-256 checksum How to use checksums |
3499a427f9c6ef1033c5ca89e9e0d85119371d6cfdf29667a97a6de33d0dd3bf
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 25, 2026.
Transparency log