Vulnerability Research|Third Rebrand Analysis

OpenClaw Security Analysis: Same Vulnerabilities, Third Name

OpenClaw (formerly Clawdbot, then Moltbot) rebranded again on January 30, 2026. Our security assessment found 900+ exposed instances, 8 critical vulnerabilities, and 181 leaked secrets - the same issues we documented in Moltbot remain unfixed. The third shell-shedding doesn't change the security posture.

This report analyzes OpenClaw, the latest rebrand of the viral AI assistant formerly known as Clawdbot and Moltbot. Despite 108,000+ GitHub stars and 15,200+ forks, the project continues to expose users to critical authentication bypass (CVE-2025-49596), remote code execution (CVE-2025-6514), and credential theft vulnerabilities. Infostealers like RedLine and Lumma have already updated to target OpenClaw installations.

Sable Security Research Team
January 30, 2026
0 views
15 min read

TL;DR: OpenClaw = Moltbot = Clawdbot (Same Vulnerabilities)

900+
Instances Exposed
via Shodan (Jan 2026)
512
Security Findings
from Argus audit
8
Critical Vulns
CVSS 7.5-9.8
181
Secrets Leaked
65 still valid

Key finding: OpenClaw is the third name for the same codebase. All vulnerabilities we reported in Moltbot remain present. The rebrand was for trademark reasons, not security.

The Rebrand Timeline: Clawdbot to Moltbot to OpenClaw

Three names in three months. Each rebrand was triggered by external pressure - never by security concerns:

Nov 2025
Clawdbot
Original release by Peter Steinberger
TRADEMARK
Jan 2026
Moltbot
Renamed after Anthropic trademark concerns
NO FIXES
Jan 30, 2026
OpenClawCURRENT
Rebranded again - same vulnerabilities remain
SAME CODE
Name Changes Don't Fix Code

Peter Steinberger acknowledged the name changes were due to trademark concerns from Anthropic. The security vulnerabilities that expose 900+ instances remain unchanged across all versions.

Industry Expert Warnings About OpenClaw/Moltbot

Don't run Clawdbot.

Heather Adkins
VP Google Cloud

How could someone trust that thing with full system access?

Yassine Aboukir
Principal Security Consultant

8 Critical Vulnerabilities in OpenClaw (Unchanged from Moltbot)

The Argus Security audit identified 512 findings. These 8 critical vulnerabilities enable full system compromise from the network without authentication:

Gateway Auth BypassCVE-2025-49596
CVSS 9.8

Reverse proxy makes all connections appear as localhost

mcp-remote RCECVE-2025-6514
CVSS 9.6

Command injection via OAuth endpoint in MCP servers

Plaintext OAuth Storage
CVSS 9.1

Tokens stored unencrypted at ~/.openclaw/credentials/

Webhook Signature Bypass
CVSS 8.8

skipVerification option allows complete bypass

Hardcoded OAuth Secrets
CVSS 8.6

Base64-encoded client secrets in source code

Path Traversal
CVSS 8.4

Agent directory paths not validated

CSRF in OAuth Flows
CVSS 8.1

State validation uses flawed nullish coalescing

Token Refresh Race
CVSS 7.5

File-based locking fails causing race conditions

CVE-2025-49596: The Localhost Trust Assumption

The root cause remains the same flawed assumption: "If you can reach localhost, you're already on the machine." In reality, Docker networks, reverse proxies, and cloud deployments routinely expose localhost to the internet.

message-handler.js - Still vulnerable in OpenClaw
// The vulnerability that exposed everything - STILL PRESENT
const isLocalClient = !hasUntrustedProxyHeaders && isLocalGatewayAddress(clientIp);
const nonceRequired = !isLocalClient;// localhost = no auth required
// Attack flow - unchanged in OpenClaw
┌──────────────┐     ┌─────────────────┐     ┌─────────────────┐
│   Attacker   │ ──► │ nginx/Caddy     │ ──► │ OpenClaw Gateway│
│  (Internet)  │     │ (Reverse Proxy) │     │  (Port 8080)    │
└──────────────┘     └─────────────────┘     └─────────────────┘
                              │
                    clientIp = 127.0.0.1
                    isLocalClient = true
                    nonceRequired = false
                              │
                    ┌─────────▼─────────┐
                    │   FULL ACCESS      │
                    │   No auth needed  │
                    └───────────────────┘

Exposed API Endpoints in OpenClaw Instances

When authentication is bypassed, these endpoints become accessible to any attacker:

EndpointRiskExposed Data
/api/configCRITICALFull configuration with API keys, tokens, secrets
/api/agentsCRITICALAgent list, settings, and auth profiles
/api/toolsHIGHAvailable tools and execution permissions
/gateway/statusHIGHGateway status and authentication tokens
/api/sessionsHIGHSession transcripts and conversation logs

Plaintext Credential Storage Locations

OpenClaw stores all credentials in plaintext JSON files. 181 unique secrets have leaked to public repositories, with 65 still valid at time of detection:

~/.openclaw/clawdbot.json
Gateway tokens
CRITICAL
~/.openclaw/credentials/oauth.json
OAuth access/refresh tokens
CRITICAL
~/.openclaw/identity/device-auth.json
Device authentication
HIGH
~/.openclaw/credentials/whatsapp/*/creds.json
WhatsApp credentials
CRITICAL
~/.openclaw/agents/*/agent/auth-profiles.json
Model provider auth
CRITICAL
~/.openclaw/agents/*/sessions/*.jsonl
Conversation transcripts
HIGH

Infostealers Now Targeting OpenClaw Installations

Major malware families updated their targeting within 24 hours of OpenClaw's release:

RedLine

Added OpenClaw directory enumeration

Black market: $50-200
Lumma

Targeting credential JSON files

Black market: $30-150
Vidar

Harvesting conversation logs

Black market: $20-50

MITRE ATT&CK Mapping

TechniqueNameOpenClaw Relevance
T1133External Remote ServicesExposed gateway instances
T1210Exploitation of Remote ServicesCVE-2025-49596 auth bypass
T1552.001Credentials In FilesPlaintext credential storage
T1574Hijack Execution FlowSkill/plugin poisoning
T1659Content InjectionPrompt injection attacks
T1195.002Supply Chain CompromiseClawdHub malicious skills

Minimum Secure Configuration for OpenClaw

If you must run OpenClaw, this is the minimum configuration to prevent unauthenticated access:

openclaw.yaml - Hardened configuration
gateway:
  bind: "127.0.0.1:8080"
  auth:
    token: "${OPENCLAW_AUTH_TOKEN}"
  trustedProxies:
    - "10.0.0.0/8"

agents:
  defaults:
    sandbox:
      enabled: true
      scope: "session"
    dm:
      policy: "pairing"

tools:
  elevated:
    allowFrom: []

Immediate Remediation Steps

If you run OpenClaw, Moltbot, or Clawdbot, take these steps immediately:

1Rotate ALL exposed secrets
# 181 secrets leaked - 65 still valid
# Rotate immediately:
- LLM API keys (OpenAI, Anthropic)
- OAuth tokens (Google, Slack, Discord)
- Bot tokens (Telegram, WhatsApp)
- SSH keys if exposed
2Enable gateway authentication
gateway:
  auth:
    token: "$(openssl rand -hex 32)"
3Configure trustedProxies
gateway:
  trustedProxies:
    - "10.0.0.0/8"
    - "172.16.0.0/12"
    - "192.168.0.0/16"
4Update mcp-remote to 0.1.16+
npm update mcp-remote@latest
# Fixes CVE-2025-6514 (CVSS 9.6)
5Update Node.js to 22.12.0+
# Fixes:
# CVE-2025-59466 (async_hooks DoS)
# CVE-2026-21636 (permission bypass)

Related Research

Is Your AI Infrastructure Secure?

OpenClaw is just one of many AI tools with security issues. We audit AI agent deployments, LLM integrations, and agentic infrastructure before attackers find your vulnerabilities.

This research was conducted following responsible disclosure practices. All testing was limited to benign enumeration. No user data was accessed or stored.