# GHSA-2CCW-8RVH-6CMG: @fastify/reply-from vulnerable to path traversal via scheme-prefixed encoded dot-segments

- Database: GitHub
- Severity: CVSS 8.2 (High)
- Published: 2026-11-19
- Modified: 2026-11-19

## Tracked as

- [CVE-2026-130886](https://cve.now/CVE-2026-130886.md): Fastify Reply From path traversal in request forwarding

## Affected versions

- @fastify/reply-from (npm): from 12.8.1, fixed in 12.8.4

## Description

### Impact

`@fastify/reply-from`'s `buildURL()` validated the attacker-supplied `source` string for `.` and `..` segments, but when `source` begins with the same special scheme as the base (`http:` or `https:`), the WHATWG `URL` parser drops the scheme and resolves the rest as a relative path. A payload such as `http:%2e%2e%2fsecret.txt` therefore passed the check (the segment glued to the scheme read as `http:..`, not `..`) while the proxy forwarded `%2e%2e%2fsecret.txt`. An upstream that percent-decodes the request path before resolving dot-segments then serves a path outside the configured prefix. An application that scopes proxying to an upstream prefix (`reply.from('/v1/public/' + req.params['*'])`), or relies on Fastify route guards in front of `reply.from()`, can have an unauthenticated request escape the intended prefix and reach sibling or parent upstream endpoints. Releases 12.8.1 through 12.8.3 are affected: the per-segment check introduced in 12.8.1 is what the scheme prefix bypasses, and up to 12.8.0 any decoded `../` was rejected.

### Patches

Upgrade to `@fastify/reply-from` 12.8.4. `buildURL()` now also checks for dot-segments on the decoded path of the parsed URL, which is the path actually sent upstream, so a scheme prefix can no longer change what gets validated.

### Workarounds

None.

### Out of scope

An upstream that decodes the request path more than once (for example `unquote(unquote(path))`) can still resolve a double-encoded payload such as `%252e%252e%252f` to a traversal, independent of any proxy in front of it. That is an upstream path-traversal issue (CWE-174) and should be fixed there.

## References

- https://github.com/fastify/fastify-reply-from/security/advisories/GHSA-2ccw-8rvh-6cmg

Page: https://cve.now/GHSA-2CCW-8RVH-6CMG
