CVEReports
CVEReports

Automated vulnerability intelligence platform. Comprehensive reports for high-severity CVEs generated by AI.

Product

  • Home
  • Sitemap
  • RSS Feed

Company

  • About
  • Contact
  • Privacy Policy
  • Terms of Service

© 2026 CVEReports. All rights reserved.

Made with love by Amit Schendel & Alon Barad



GHSA-8MXV-9XHP-86H4

GHSA-8MXV-9XHP-86H4: Information Disclosure and Credential Leakage during S3 HTTP Redirects in rclone

Amit Schendel
Amit Schendel
Senior Security Researcher

Aug 6, 2026·7 min read·13 visits

Executive Summary (TL;DR)

Rclone versions prior to 1.75.0 leak sensitive authentication headers and SSE-C customer-provided encryption keys during protocol downgrades and cross-host HTTP redirects.

A critical security flaw was identified in rclone before version 1.75.0, where the custom S3 redirect handler failed to sanitize sensitive authentication headers and encryption keys during cross-host redirects or transport downgrades. This flaw allows attackers on the path or controlling target hosts to intercept sensitive IBM IAM tokens, AWS S3 Express tokens, and customer-provided server-side encryption keys (SSE-C).

Vulnerability Overview

The S3 backend component in rclone is responsible for orchestrating file transfers, access queries, and endpoint communications with Amazon Simple Storage Service (S3) and compliant object storage platforms. As object storage architectures scale dynamically, S3 endpoints frequently utilize HTTP redirection messages (such as 301, 302, 307, or 308 response codes) to direct clients to specific regional endpoints, caching layers, or partition nodes. Because of these redirects, rclone maintains a custom redirect evaluation module to control the behavior of subsequent connections.\n\nA critical security vulnerability exists in the redirect logic of rclone versions prior to 1.75.0. When following redirects, rclone's handler fails to thoroughly cleanse sensitive headers before dispatching the subsequent request. This leads to information disclosure where critical cloud credentials, session identifiers, and cryptographic keys are transmitted either over plaintext channels or exposed directly to unauthorized target hosts. This vulnerability is classified under CWE-200 (Exposure of Sensitive Information to an Unauthorized Actor).\n\nThe attack surface is exposed whenever rclone is configured to interface with an S3-compatible service that issues redirection payloads, or when an adversary is positioned to influence the transport routing of the active session. This technical analysis provides an exhaustive breakdown of the root cause, execution mechanics, patch structure, and defensive recommendations.

Root Cause Analysis

The fundamental cause of the vulnerability lies in the default handling of HTTP redirection within the Go standard library's net/http client, combined with an incomplete sanitization blocklist within rclone's custom redirect handler. Go's standard library implements an automated security measure that strips the conventional Authorization header during cross-host redirects. However, this library mechanism possesses two critical operational gaps that leave custom and protocol-specific variables exposed.\n\nFirst, Go's default redirect policy does not strip the Authorization header when a redirect preserves the exact hostname but downgrades the transport scheme from HTTPS to HTTP. In this scenario, the standard library assumes the administrative domain remains uniform and retains the bearer token, allowing it to be transmitted unencrypted. Consequently, if a service like IBM Cloud Object Storage uses standard Authorization: Bearer <token> headers, these tokens are sent in plaintext during a scheme downgrade.\n\nSecond, modern S3 integrations rely heavily on custom HTTP headers to pass ephemeral credentials, session states, and customer-provided encryption keys. Since Go's net/http package treats custom headers as generic transport metadata, it propagates them across host boundaries unconditionally. Because rclone's original s3CheckRedirect implementation only targeted the single AWS STS header X-Amz-Security-Token for removal, all other critical headers were forwarded to any third-party domain specified in the redirect response.

Code Analysis

Analyzing the vulnerable code path in rclone's S3 backend reveals the limited safety measures in place before the fix. The function s3CheckRedirect in backend/s3/s3.go handled the verification of redirections using the following block:\n\ngo\nfunc s3CheckRedirect(req *http.Request, via []*http.Request) error {\n\tif len(via) >= 10 {\n\t return errors.New(\"stopped after 10 redirects\")\n\t}\n\tif s3RedirectCrossesHost(req, via) {\n\t req.Header.Del(\"X-Amz-Security-Token\")\n\t}\n\treturn nil\n}\n\n\nThis function fails on multiple fronts. First, it makes no effort to evaluate protocol transitions, allowing HTTPS-to-HTTP transport scheme downgrades without restriction. Second, by calling req.Header.Del("X-Amz-Security-Token") exclusively, it allows every other custom S3 and IAM header to remain in the request structure as it transfers to the new host.\n\nTo address this, the maintainers implemented patch 9328763d1b73db71e97c0332b19e3747abeb9191 and subsequent patch 7543a7a87884aca957590b20b0714078d51af87b. The modified codebase establishes an explicit blocklist of sensitive headers (s3RedirectSecretHeaders) and enforces a strict protocol downgrade block:\n\ngo\n// s3RedirectSecretHeaders are the request headers carrying origin-bound\n// secrets that must not be forwarded when a redirect crosses a host or\n// downgrades the scheme.\nvar s3RedirectSecretHeaders = []string{\n\t\"X-Amz-Security-Token\", // AWS STS session token\n\t\"X-Amz-S3session-Token\", // S3 Express session token\n\t\"Authorization\", // IBM IAM bearer token\n\t\"ibm-service-instance-id\",\n\t\"X-Amz-Server-Side-Encryption-Customer-Algorithm\",\n\t\"X-Amz-Server-Side-Encryption-Customer-Key\",\n\t\"X-Amz-Server-Side-Encryption-Customer-Key-Md5\",\n\t\"X-Amz-Copy-Source-Server-Side-Encryption-Customer-Algorithm\",\n\t\"X-Amz-Copy-Source-Server-Side-Encryption-Customer-Key\",\n\t\"X-Amz-Copy-Source-Server-Side-Encryption-Customer-Key-Md5\",\n\t\"Referer\", // May contain a presigned URL with cryptographic signatures\n}\n\n\nWith the secret header collection defined, the updated s3CheckRedirect routine performs systematic verification:\n\ngo\nfunc s3CheckRedirect(req *http.Request, via []*http.Request) error {\n\tif len(via) >= 10 {\n\t return errors.New(\"stopped after 10 redirects\")\n\t}\n\t// Never follow a redirect that downgrades the transport from HTTPS to HTTP.\n\tif via[len(via)-1].URL.Scheme == \"https\" && req.URL.Scheme == \"http\" {\n\t return fmt.Errorf(\"refusing to follow insecure redirect from HTTPS to HTTP: %s\", req.URL.Redacted())\n\t}\n\tif s3RedirectCrossesHost(req, via) {\n\t for _, header := range s3RedirectSecretHeaders {\n\t req.Header.Del(header)\n\t }\n\t}\n\treturn nil\n}\n\n\nThis multi-stage control guarantees that transport-level downgrades trigger an immediate failure, while cross-host redirections systematically strip all recognized authentication, token, reference, and cryptographic key variables.

Exploitation Scenarios

To conceptualize the exploitation vector, consider a scenario involving an Adversary-in-the-Middle (AiTM) positioned on an adjacent network or an untrusted transit path. The target system executes an automated backup task utilizing rclone and an S3-compatible backend such as IBM Cloud Object Storage. The rclone command is initiated with credentials that map to an active IBM IAM bearer token. Below is the workflow mapping the forced protocol downgrade:\n\nmermaid\nsequenceDiagram\n\tautonumber\n\tactor Client as rclone Client\n\tparticipant Attacker as On-Path Attacker\n\tparticipant S3 as S3 Endpoint\n\tClient->>Attacker: HTTPS Connection Request\n\tAttacker->>S3: Relayed HTTPS Request\n\tS3->>Attacker: 302 Redirect to same host (HTTP scheme)\n\tAttacker->>Client: Relayed Redirect (http://s3.ibm.com/path)\n\tClient->>Attacker: Plaintext HTTP Request with Authorization Header\n\tNote over Attacker: Attacker captures plain-text IBM IAM token\n\n\nBecause the host remains unchanged, the client proceeds to transmit the original request over unencrypted HTTP. The on-path attacker captures the plaintext headers, extracting the highly privileged IAM token.\n\nIn a second scenario, an attacker controls or compromises a downstream S3-compatible service. The client utilizes Server-Side Encryption with Customer-Provided Keys (SSE-C) to perform secure read or write actions. Upon receiving the initial request, the compromised endpoint responds with a 307 Temporary Redirect pointing to a malicious, attacker-controlled domain. Rclone processes this redirection, stripping X-Amz-Security-Token but preserving the X-Amz-Server-Side-Encryption-Customer-Key header, delivering the raw decryption keys directly to the attacker.

Technical Impact & Severity Assessment

The exposure of authorization tokens and encryption keys represents a high-severity confidentiality threat. An attacker who successfully intercepts an IBM IAM bearer token gains access to the associated cloud environment with the privileges of the victim's service account. This access allows unauthorized read, write, or deletion operations depending on the bucket policies.\n\nFor S3 Express (directory buckets), the exposure of the X-Amz-S3session-Token allows the attacker to authenticate directly to directory buckets within the active session window. Furthermore, the leakage of SSE-C keys compromises the confidentiality of stored data. If an attacker possesses both the ciphertext and the raw customer-provided key, they can decrypt the target object without authorization.\n\nThe CVSS 3.1 rating is calculated as 5.3 (Medium) with the vector CVSS:3.1/AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N. While the impact on confidentiality is high, the complexity associated with routing manipulation or on-path interception reduces the likelihood of random, broad-spectrum exploitation.

Remediation & Mitigation Guidance

The definitive mitigation for GHSA-8MXV-9XHP-86H4 is upgrading rclone installations to version 1.75.0 or later. This release enforces the secure protocol validation block and fully cleanses the expanded blocklist of secret headers. It is critical to audit and update automated orchestration platforms, container environments, and system processes executing rclone commands.\n\nIn scenarios where immediate software package deployment is constrained by change management policies, organizations must implement defensive configurations to reduce the attack surface. Egress firewall policies should be constructed on host machines to block outbound TCP port 80 (HTTP) traffic generated by the rclone binary, forcing the client to only maintain port 443 (HTTPS) sessions. Furthermore, administrators must verify that S3 backend endpoint configurations are explicitly hardcoded to use https:// schemas inside the rclone.conf config file.\n\n> [!NOTE]\n> Following the deployment of patched rclone binaries, security teams should assume any potentially exposed credentials are compromised and initiate a comprehensive key and token rotation cycle. This includes rotating active IBM Cloud API keys, regenerating S3 access keys, and provisioning new customer keys for SSE-C-enabled datasets.

Fix Analysis (2)

Technical Appendix

CVSS Score
5.3/ 10
CVSS:3.1/AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N

Affected Systems

rclone S3 Backend

Affected Versions Detail

Product
Affected Versions
Fixed Version
github.com/rclone/rclone
rclone
< 1.75.01.75.0
AttributeDetail
CWE IDCWE-200 / CWE-201
Attack VectorAdjacent Network
CVSS Score5.3 (Medium)
Exploit StatusPoC / Conceptual
Affected ComponentS3 backend (s3CheckRedirect)
Fixed Versionv1.75.0

MITRE ATT&CK Mapping

T1557Adversary-in-the-Middle
Credential Access
T1552Unsecured Credentials
Credential Access
CWE-200
Exposure of Sensitive Information to an Unauthorized Actor

Vulnerability Timeline

Original fix commit for IAM and SSE-C leaks added
2026-07-15
AWS S3 Express token mitigation committed
2026-07-16
GitHub Security Advisory published
2026-08-05
Rclone version 1.75.0 released with patches
2026-08-05

References & Sources

  • [1]GitHub Security Advisory GHSA-8MXV-9XHP-86H4
  • [2]rclone v1.75.0 Release Notes

Attack Flow Diagram

Press enter or space to select a node. You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.
Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

More Reports

•about 2 hours ago•CVE-2026-9639
6.5

CVE-2026-9639: Authenticated Denial of Service via Nil-Pointer Dereference in Canonical LXD Custom Volume Backup Import

CVE-2026-9639 (GHSA-j93m-3j9p-m5m8) is a null pointer dereference vulnerability in Canonical LXD affecting versions 5.0.0 through 5.21.4 and 6.0 through 6.8. The flaw resides in the (*lxdBackend).CreateCustomVolumeFromBackup method within lxd/storage/backend_lxd.go. An authenticated attacker with the can_create_storage_volumes permission can upload a crafted custom volume backup tarball that omits the expires_at field on a snapshot entry, triggering an unhandled SIGSEGV panic and terminating the LXD daemon.

Amit Schendel
Amit Schendel
5 views•9 min read
•about 3 hours ago•CVE-2026-33050
8.8

CVE-2026-33050: OS Command Injection in GeoNode Vector Upload Handler

An operating system command injection vulnerability (CWE-78) exists in GeoNode versions prior to 5.1.0. The vulnerability occurs in the vector upload processing pipeline when handling user-supplied geospatial files. By supplying a crafted layer name containing shell metacharacters, an authenticated attacker with basic data upload privileges can execute arbitrary commands on the underlying server or container hosting the GeoNode application and Celery worker processes.

Alon Barad
Alon Barad
5 views•9 min read
•about 4 hours ago•CVE-2026-46429
5.5

CVE-2026-46429: Unrestricted File Upload and Validation Bypass in GeoNode

GeoNode versions prior to 5.0.3 contain an unrestricted file upload vulnerability (CWE-434) within the document and dataset ingestion architecture. Authenticated low-privileged users can bypass upload validation via direct REST API interactions, file extension masquerading, metadata mutation via PATCH requests, or malicious archive submissions. This allows attackers to store and serve dangerous file types, alter on-disk file references, or trigger path traversal and resource exhaustion during archive extraction.

Amit Schendel
Amit Schendel
6 views•10 min read
•about 5 hours ago•CVE-2026-55649
6.3

CVE-2026-55649: Stored Cross-Site Scripting in GeoNode Document Metadata Abstract Field

CVE-2026-55649 is a stored Cross-Site Scripting (XSS) vulnerability in GeoNode versions prior to 5.0.3. The flaw exists because document abstract metadata fields are ingested via the metadata management pipeline and stored in the backend database without HTML sanitization or input neutralization. An authenticated attacker with privileges to create or edit document metadata can inject arbitrary HTML and JavaScript payloads into these fields. When another user, such as an administrator, views the affected document page or catalog listing, the unencoded payload is rendered into the Document Object Model (DOM) and executes within the victim's browser session.

Amit Schendel
Amit Schendel
7 views•9 min read
•about 6 hours ago•GHSA-QX75-2P3R-PWM5
7.7

CVE-2026-55622: Cross-Project Instance Copy Authorization Bypass in Canonical LXD and LXC Incus

CVE-2026-55622 (GHSA-qx75-2p3r-pwm5) is an improper access control vulnerability (CWE-284) in Canonical LXD and LXC Incus (tracked as GHSA-c9f5-j9c3-mhrg). In multi-tenant deployments utilizing project-based isolation, the REST API handler for instance creation via POST /1.0/instances verifies that the authenticated user has authorization in the destination project specified by the query parameter, but fails to verify whether the user holds read or view permissions for the source instance when executing a copy operation across projects. Consequently, an authenticated client restricted to a single project can clone instances or snapshots from unauthorized projects into their own project, resulting in complete disclosure of the target instance filesystem and embedded credentials.

Amit Schendel
Amit Schendel
6 views•11 min read
•about 7 hours ago•GHSA-7MR3-28H5-M5VX
7.7

GHSA-7mr3-28h5-m5vx: Authorization Bypass and Cross-Project Volume Copy in Canonical LXD

An authorization bypass vulnerability in Canonical LXD (tracked as GHSA-7mr3-28h5-m5vx and CVE-2026-55621, and as GHSA-64f3-v33m-w89f in Linux Containers Incus) allows an authenticated user with restricted project access to copy custom storage volumes and volume snapshots from unauthorized projects. The flaw resides in the storagePoolVolumesPost REST API handler, which verifies permissions on the destination project but fails to check whether the caller holds view entitlements on the source project or volume specified in the request body.

Alon Barad
Alon Barad
5 views•10 min read