CVE-2026-35656: Openclaw Auth Bypass Vulnerability

CVE-2026-35656 Overview

CVE-2026-35656 is an authentication bypass vulnerability in OpenClaw, a Node.js application, affecting versions prior to 2026.3.22. The vulnerability exists in the X-Forwarded-For header processing when trustedProxies is configured, allowing remote attackers to spoof loopback hops and bypass security controls. By injecting forged forwarding headers, attackers can masquerade as loopback clients to circumvent canvas authentication and rate-limiting protections.

Critical Impact

Remote attackers can bypass authentication and rate-limiting controls by spoofing X-Forwarded-For headers, potentially gaining unauthorized access to protected resources and evading abuse prevention mechanisms.

Affected Products

• OpenClaw versions prior to 2026.3.22
• OpenClaw for Node.js with trustedProxies configuration enabled
• Deployments behind reverse proxies or load balancers

Discovery Timeline

• 2026-04-10 - CVE-2026-35656 published to NVD
• 2026-04-13 - Last updated in NVD database

Technical Details for CVE-2026-35656

Vulnerability Analysis

This authentication bypass vulnerability (CWE-290: Authentication Bypass by Spoofing) affects OpenClaw's proxy trust mechanism. When the trustedProxies configuration is enabled, the application relies on X-Forwarded-For headers to determine the originating client IP address. However, insufficient validation of these headers allows attackers to inject arbitrary IP addresses, including loopback addresses (e.g., 127.0.0.1 or ::1).

The flaw resides in how OpenClaw processes forwarding headers in the authentication and rate-limiting subsystems. When a request appears to originate from a loopback address, the application may grant elevated trust levels, bypassing authentication checks or rate limits designed to protect against abuse.

Root Cause

The root cause is improper validation of the X-Forwarded-For header chain when trustedProxies is configured. The application fails to properly verify that loopback addresses in the forwarding chain were actually added by trusted proxy hops rather than injected by malicious clients. This allows attackers to prepend or inject loopback IP addresses into the header, causing the server to incorrectly identify the request as originating from a trusted local source.

Attack Vector

The attack is network-based and requires no authentication or user interaction. An attacker can craft HTTP requests with manipulated X-Forwarded-For headers to exploit this vulnerability:

1. The attacker sends a request to the OpenClaw application with a forged X-Forwarded-For header containing a loopback address
2. The application's proxy trust logic processes the header and extracts the spoofed loopback IP
3. Canvas authentication and rate-limiting components treat the request as originating from localhost
4. Authentication requirements and rate limits are bypassed, granting the attacker unauthorized access

The vulnerability mechanism involves manipulation of HTTP forwarding headers. When trustedProxies is enabled, the application parses the X-Forwarded-For header to determine client identity. Attackers can inject loopback addresses (such as 127.0.0.1) into this header chain. The authentication and rate-limiting logic then incorrectly identifies the request as coming from a local, trusted source. For detailed technical analysis, see the GitHub Security Advisory.

Detection Methods for CVE-2026-35656

Indicators of Compromise

• Unusual authentication bypass patterns from external IP addresses with loopback addresses in X-Forwarded-For headers
• Rate limit violations or excessive requests that appear to originate from 127.0.0.1 or ::1 but trace back to external sources
• Authentication logs showing successful canvas authentication from inconsistent IP address chains
• Web server access logs containing suspicious X-Forwarded-For header values with loopback addresses

Detection Strategies

• Monitor and alert on X-Forwarded-For headers containing loopback addresses (127.0.0.1, ::1, localhost) from external connections
• Implement log analysis rules to detect authentication successes correlated with forged forwarding headers
• Deploy web application firewall (WAF) rules to inspect and validate X-Forwarded-For header chains
• Enable detailed logging of all proxy-related headers for forensic analysis

Monitoring Recommendations

• Configure SIEM alerts for anomalous patterns where external requests contain internal/loopback IP addresses in forwarding headers
• Establish baseline metrics for canvas authentication patterns and alert on statistical anomalies
• Monitor rate-limiting bypass attempts by correlating request volumes with apparent source IPs
• Review authentication logs regularly for requests that bypass normal authentication flows

How to Mitigate CVE-2026-35656

Immediate Actions Required

• Upgrade OpenClaw to version 2026.3.22 or later immediately
• Review trustedProxies configuration and ensure only legitimate proxy IP addresses are whitelisted
• Audit access logs for evidence of past exploitation attempts
• Consider temporarily disabling trustedProxies if upgrade is not immediately possible

Patch Information

OpenClaw has released security patches addressing this vulnerability. The fixes are available in the following commits:

• Commit 630f1479c44f78484dfa21bb407cbe6f171dac87
• Commit fc2d29ea926f47c428c556e92ec981441228d2a4

For additional details, refer to the GitHub Security Advisory GHSA-844j-xrrq-wgh4 and the VulnCheck Advisory.

Workarounds

• Restrict trustedProxies configuration to explicitly enumerate trusted proxy IP addresses rather than using broad patterns
• Implement network-level controls to strip or validate X-Forwarded-For headers at the edge before requests reach OpenClaw
• Deploy a reverse proxy or WAF that performs strict validation of forwarding headers and rejects requests with loopback addresses from external sources
• Consider implementing additional authentication factors that do not rely on IP-based trust

# Example: Configure trusted proxies with explicit IP addresses only
# In OpenClaw configuration, avoid wildcard proxy trust
# Instead of trustedProxies: true or trustedProxies: "*"
# Use explicit trusted proxy IPs:
# trustedProxies: ["10.0.0.1", "10.0.0.2"]

# Nginx example: Strip X-Forwarded-For from untrusted sources
# Add to nginx configuration before proxying to OpenClaw:
# proxy_set_header X-Forwarded-For $remote_addr;