SearXNGSAFE
Private web search for AI assistants via SearXNG — supports Claude, Cursor, and any MCP client
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Privacy-respecting web search for AI assistants — use an operator-controlled or trusted SearXNG instance with Claude, Cursor, and more.
[](https://github.com/ihor-sokoliuk/mcp-searxng/stargazers) [](https://www.npmjs.com/package/mcp-searxng) [](https://www.npmjs.com/package/mcp-searxng) [](https://hub.docker.com/r/isokoliuk/mcp-searxng) [](LICENSE) [](https://scorecard.dev/viewer/?uri=github.com/ihor-sokoliuk/mcp-searxng) [](https://www.bestpractices.dev/projects/13143) [](https://glama.ai/mcp/servers/ihor-sokoliuk/mcp-searxng) [](https://github.com/mcp/ihor-sokoliuk/mcp-searxng)
An MCP server that integrates the SearXNG API, giving AI assistants web search capabilities.
✨ Featured in the GitHub MCP Registry.
Quick Start
You need an existing SearXNG instance with JSON search enabled. This project connects an MCP client to SearXNG; it does not install SearXNG. Use an instance you operate or
9f1716c8a2dbOBSERVED · 2026-09-25Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add mcp-searxng --env SEARXNG_URL=${SEARXNG_URL} --env AUTH_USERNAME=${AUTH_USERNAME} --env AUTH_PASSWORD=${AUTH_PASSWORD} -- npx -y [email protected]Exposed tools (4)
4 read · 0 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
searxng_instance_info | read | Discovers capabilities from all reachable configured SearXNG instances via /config, including categories.common/available, engines.common/available, defaults, locales, and plugins. |
searxng_search_suggestions | read | Returns autocomplete suggestions from the configured SearXNG instance. |
searxng_web_search | read | Searches the web using SearXNG and returns a list of results, each with a title, URL, and content snippet. |
web_url_read | read | Fetches a URL and returns readable content as markdown. |
Trust audit
SAFEgrade B · trust 86/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (23)
'JVBERi0xLjQKMSAwIG9iago8PCAvVHlwZSAvQ2F0YWxvZyAvUGFnZXMgMiAwIFIgPj4KZW5kb2JqCjIgMCBvYmoKPDwgL1R5cGUgL1BhZ2VzIC9LaWRzIFszIDAgUl0gL0NvdW50IDEgPj4KZW5kb2JqCjMgMCBvYmoKPDwgL1R5cGUgL1BhZ2UgL1BhcmVudCAyIDA
.env.e2e.example
import { packageVersion } from '../../src/version.js';import { isWebUrlReadArgs } from "../../src/index.js";import { isSearXNGWebSearchArgs } from "../../src/types.js";import { searchCache } from '../../src/search-cache.js';import { setSearxngFetchForTesting } from '../../src/proxy.js';- **DNS-resolved private-address SSRF in `web_url_read` blocked (GHSA-mrvx-jmjw-vggc):** The URL reader previously validated only the literal hostname string, so a public-looking hostname that DNS-res
'metadata.example': [{ address: '169.254.169.254', family: 4 }],assert.equal(isPrivateIpv4('169.254.169.254'), true);'0.0.0.0', '10.0.0.1', '100.64.0.1', '127.0.0.1', '169.254.169.254',
Every present `Origin` on `/mcp` is validated in all modes; an absent `Origin` remains valid for non-browser clients. In non-hardened mode, an unset `MCP_HTTP_ALLOWED_ORIGINS` defaults to the exact HT
When an `/mcp` request includes an `Origin`, the server validates it against the applicable allowlist in hardened and non-hardened modes; requests without `Origin` follow the accepted non-browser path
url: `http://127.0.0.1:${address.port}`,SEARXNG_URL: 'http://127.0.0.1:1',
return { child, url: new URL(`http://127.0.0.1:${port}/mcp`), output, close: createChildCloser(child, output) };"JVBERi0xLjMKJeLjz9MKMSAwIG9iago8PAovUHJvZHVjZXIgPDM1YzY5Y2I1ZTA+Cj4+CmVuZG9iagoyIDAgb2JqCjw8Ci9UeXBlIC9QYWdlcwovQ291bnQgMQovS2lkcyBbIDQgMCBSIF0KPj4KZW5kb2JqCjMgMCBvYmoKPDwKL1R5cGUgL0NhdGFsb2cKL1BhZ2V
@types/cors, @types/express, cors, express, node-html-markdown, @types/supertest, cross-env, eslint-plugin-security
- **Modern MCP protocol serving with the official split SDK v2 packages:** HTTP and STDIO now support the modern `2026-07-28` protocol while retaining the documented legacy transports, tool and resour
- **Bounded SearXNG response-body consumption:** Search JSON and HTML fallback, `/config`, and suggestions now share a streaming response reader with a configurable `SEARXNG_MAX_RESPONSE_BYTES` ceilin
- **Established HTTP sessions now receive the configured session rate limit:** Each `POST /mcp` request now passes through exactly one limiter. Requests with a currently live `mcp-session-id` use the
In stateless mode, every POST creates a fresh MCP server and transport, ignores incoming `mcp-session-id` headers, and never emits a response session ID. A POST can return negotiated JSON or an SSE st
Rate limiting is always active in HTTP mode to prevent resource exhaustion. Before the MCP handler runs, each request is counted by resolved client IP against exactly one limit. In stateful mode, reta
Gates applied: no_behavioural_pass.
9f1716c8a2dbfull audit observations/trust-audit/mcp-server/ihor-sokoliuk__searxng.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-09-25 | 9f1716c8a2db | SAFE | B | 86 | first audit |
Questions
What is the SearXNG MCP server?
Private web search for AI assistants via SearXNG — supports Claude, Cursor, and any MCP client
What tools does SearXNG expose?
4 in total: 4 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is SearXNG safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (86/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does SearXNG need?
It reads AUTH_PASSWORD, AUTH_USERNAME, MCP_HTTP_AUTH_MODE, MCP_HTTP_AUTH_TOKEN, MCP_HTTP_OAUTH_ISSUER, MCP_HTTP_OAUTH_SCOPES and SEARXNG_URL from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does SearXNG run?
It speaks stdio and streamable-http, so it runs as a local process your client starts. It is published on npm as mcp-searxng at 2.4.0.
How current is this page?
The grade is for one exact copy of the source (9f1716c8a2db), read on 2026-09-25. The repository is watched and re-audited when it changes.