Skip to content

http backend forwards custom/auth headers to a different host on redirect

Low
ncw published GHSA-486v-q2wf-fp2r Sep 4, 2026

Package

gomod github.com/rclone/rclone (Go)

Affected versions

>= 1.49.0, <= 1.75.0

Patched versions

1.75.1

Description

Vulnerability Details

File: backend/http/http.go
Lines: 285 (client construction — no CheckRedirect), 505-510 (addHeaders, writes configured secret headers onto every request), 533-534 / 700-701 / 782-785 (f.httpClient.Do(req) used by List/stat/download)
CWE: CWE-200 — Exposure of Sensitive Information to an Unauthorized Actor (also CWE-319, CWE-522)
Severity: Medium
CVSS: 5.3 — CVSS:3.1/AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N

Root Cause

The http backend lets a user attach arbitrary secret headers to every request via --http-headers/headers= (documented for authentication: '"Cookie","name=value","Authorization","xxx"'). The backend's HTTP client is built with fshttp.NewClient(ctx), which never sets http.Client.CheckRedirect, so it falls back to Go's stdlib default redirect policy.

Go's default policy only strips four header names (Authorization, Www-Authenticate, Cookie, Cookie2), and only when the redirect target's host differs from the original — every other configured header is copied to the redirect target unconditionally, regardless of host or scheme. Even the four protected names survive a same-host https:// → http:// downgrade, since Go only checks host equality, not scheme.

Any redirect response from the configured remote — whether from server compromise, an open redirect, a CDN/mirror failover to a different domain, or a malicious server from the start — causes rclone to resend every configured secret header (and, for a scheme downgrade, Authorization/Cookie in cleartext) to the new destination.

This is the exact vulnerability class already fixed for the s3 backend (9328763/7543a7a, GHSA-8mxv-9xhp-86h4 — which I originally reported as GHSA-gx4c-2hqx-cw2r) and the webdav backend (59b513b, GHSA-h4mf-4v27-hggj, wiring rest.RefuseHTTPSDowngradeRedirectFn). backend/http was not touched by either fix.

Vulnerable Code

// backend/http/http.go:285
client := fshttp.NewClient(ctx)   // no CheckRedirect set
...
f.httpClient = client             // used by readDir / NewObject / Object.Open
// backend/http/http.go:505-510
func addHeaders(req *http.Request, opt *Options) {
	for i := 0; i < len(opt.Headers); i += 2 {
		key := opt.Headers[i]
		value := opt.Headers[i+1]
		req.Header.Add(key, value)
	}
}

Attack Scenario

  1. User configures an http remote: url=https://good.example.com/files/, headers=X-Api-Key,SECRET-TOKEN.
  2. At some point good.example.com returns a redirect whose Location points at a different host (compromise, open redirect, CDN change, or malice from the start).
  3. User runs any operation (ls, cat, copy, mount, serve) against the remote.
  4. rclone follows the redirect with the default client and resends X-Api-Key: SECRET-TOKEN to the new, untrusted destination.
  5. The attacker's server captures the secret from the incoming request.

Impact

Exfiltration of API keys / bearer tokens / session cookies configured for one host, to any host the (trusted-at-configuration-time) remote later redirects to. All operations on the http backend (list, stat, download, mount, serve) are affected. No special rclone privileges or unusual user interaction are needed beyond a normal sync/list/copy once the redirect exists.

Dynamic Confirmation

Built rclone from source at cfdc9d0 (current master, v1.76.0-DEV) and configured:

[testhttp]
type = http
url = http://127.0.0.1:9090/
headers = X-Api-Key,SUPER-SECRET-TOKEN-abc123

Server A (port 9090, the "configured" host) 302-redirects every request to Server B (port 9091, a different host). Running rclone cat testhttp:file.txt caused Server B — which was never configured with any credential — to receive:

Header: X-Api-Key: SUPER-SECRET-TOKEN-abc123
Header: Referer: http://127.0.0.1:9090/file.txt

rclone printed Server B's response body as if it were the real file, confirming the full stat→redirect→download round trip leaks the header and trusts the redirect target.

Vulnerable Code / Fix

A minimal fix (implemented, tested, and verified to close the leak while preserving redirect functionality) wires the client to rest.RefuseHTTPSDowngradeRedirectFn (already used by webdav) and strips the configured opt.Headers on any cross-host redirect:

client := fshttp.NewClient(ctx)
client.CheckRedirect = redirectCheckFn(opt)
...
func redirectCheckFn(opt *Options) func(req *http.Request, via []*http.Request) error {
	return func(req *http.Request, via []*http.Request) error {
		if err := rest.RefuseHTTPSDowngradeRedirectFn(req, via); err != nil {
			return err
		}
		if len(via) > 0 && req.URL.Host != via[0].URL.Host {
			for i := 0; i < len(opt.Headers); i += 2 {
				req.Header.Del(opt.Headers[i])
			}
		}
		return nil
	}
}

A regression test (TestRedirectStripsHeadersOnHostChange) was added to backend/http/http_internal_test.go, confirmed to fail without the fix and pass with it. Full backend/http and lib/rest test suites pass with the fix applied. I have a fix branch ready to push to a private fork once this report is acknowledged.

Verification

Dynamically confirmed on rclone master @ cfdc9d0 (post v1.75.0) in a local test harness — see "Dynamic Confirmation" above. Fix verified to eliminate the leak via the same harness (secret header absent from Server B after the fix; functionality — file download via redirect — unaffected).

Severity

Low

CVSS overall score

This score calculates overall vulnerability severity from 0 to 10 and is based on the Common Vulnerability Scoring System (CVSS).
/ 10

CVSS v3 base metrics

Attack vector
Network
Attack complexity
High
Privileges required
None
User interaction
None
Scope
Unchanged
Confidentiality
Low
Integrity
None
Availability
None

CVSS v3 base metrics

Attack vector: More severe the more the remote (logically and physically) an attacker can be in order to exploit the vulnerability.
Attack complexity: More severe for the least complex attacks.
Privileges required: More severe if no privileges are required.
User interaction: More severe when no user interaction is required.
Scope: More severe when a scope change occurs, e.g. one vulnerable component impacts resources in components beyond its security scope.
Confidentiality: More severe when loss of data confidentiality is highest, measuring the level of data access available to an unauthorized user.
Integrity: More severe when loss of data integrity is the highest, measuring the consequence of data modification possible by an unauthorized user.
Availability: More severe when the loss of impacted component availability is highest.
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N

CVE ID

CVE-2026-88013

Weaknesses

Exposure of Sensitive Information to an Unauthorized Actor

The product exposes sensitive information to an actor that is not explicitly authorized to have access to that information. Learn more on MITRE.

Cleartext Transmission of Sensitive Information

The product transmits sensitive or security-critical data in cleartext in a communication channel that can be sniffed by unauthorized actors. Learn more on MITRE.

Insufficiently Protected Credentials

The product transmits or stores authentication credentials, but it uses an insecure method that is susceptible to unauthorized interception and/or retrieval. Learn more on MITRE.

Credits