CVE-2026-49857: auth-fetch-mcp is an MCP server that lets AI assistants fetch content from authenticated web pages. Version 3.0.1 implem…
auth-fetch-mcp is an MCP server that lets AI assistants fetch content from authenticated web pages. Version 3.0.1 implements SSRF protection in `assertSafeUrl()` (`src/security.ts`) to block requests to private and loopback addresses. However, the `isPrivateV6()` function fails to detect IPv4-mapped IPv6 loopback addresses in their hex-normalized form. When an attacker supplies a URL such as `http://[::ffff:127.0.0.1]:PORT/`, the Node.js WHATWG URL parser silently normalizes the host to `[::ffff:7f00:1]`. Because `net.isIPv4('7f00:1')` returns `false`, the private-IP check is bypassed and the URL is passed to the browser or HTTP client, allowing the MCP tool to reach loopback services that are supposed to be blocked. The issue is exploitable under default configuration without any special environment variable. Version 3.0.1 patches the issue.
- Published by source
- 2026-08-13
- First seen by Reality Radar
- 2026-10-23
- Last verified
- 2026-10-23
What the official notice says
Risk: CVSS 7.4
Is this affecting you?
Add what you own or use and Reality Radar will tell you when official changes match it — with evidence and next steps.
Sources
- NIST National Vulnerability Database Tier A · official published 2026-08-13
Evidence ledger
Every piece of evidence we store is appended to a public hash chain, so later edits are detectable. Anyone can re-verify it.
| Entry | Kind | Observed | Evidence hash | Entry hash |
|---|---|---|---|---|
| #1299 | new | 2026-10-23 | c48110eace58…42651a | d51a73a4d0ce…5f11d5 |
Chain head: #1479 ad3417dbc128…777bd4