Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
Agent must already have privileged tool access (PR:L) and low complexity (AC:L); disabling the consent boundary to gain command execution crosses a security scope (S:C) with high C/I/A impact.
Primary rating from NVD.
CVSS VectorNVD
Lifecycle Timeline
6Blast Radius
ecosystem impact- 3 npm packages depend on openclaw (3 direct, 0 indirect)
Ecosystem-wide dependent count for version 2026.3.28.
DescriptionCVE.org
OpenClaw before 2026.3.28 contains an agentic consent bypass vulnerability allowing LLM agents to silently disable execution approval via config.patch parameter. Remote attackers can exploit this to bypass security controls and execute unauthorized operations without user consent.
AnalysisAI
Authorization bypass in OpenClaw before 2026.3.28 lets an LLM agent silently disable execution-approval controls by writing to protected configuration paths (tools.exec.ask and tools.exec.security) through the gateway tool's config.patch and config.apply actions. Once the exec-approval guardrail is turned off, the agent can run commands without the user consent prompt that normally gates them, effectively self-escalating its own operational authority. EPSS is low (0.11%, 28th percentile) and there is no public exploit identified at time of analysis, but the technical impact is rated total by CISA's SSVC framework and a vendor patch is available.
Technical ContextAI
OpenClaw is an agentic runtime (published on npm as the openclaw package) in which LLM agents call gateway tools to read and mutate their own runtime configuration. Command execution is guarded by exec-approval settings - specifically tools.exec.ask (when to prompt, e.g. 'on-miss') and tools.exec.security (e.g. 'allowlist' vs 'full'). The root cause is CWE-862 (Missing Authorization): the config.patch and config.apply gateway actions applied a JSON5 merge-patch to the live config without treating these approval settings as protected, so an agent could overwrite them the same way it edits any benign key. The commit diff (76411b2) introduces a PROTECTED_GATEWAY_CONFIG_PATHS allowlist ['tools.exec.ask','tools.exec.security'], fetches the current config via config.get, and rejects any patch/apply that would alter those paths - including via the legacy tools.bash alias that maps onto tools.exec.security, and including config.apply calls that omit the protected settings entirely.
RemediationAI
Vendor-released patch: upgrade the openclaw npm package to 2026.3.28 or later (2026.3.31 is the latest published release), which enforces the PROTECTED_GATEWAY_CONFIG_PATHS check as delivered in fix commit 76411b2afc4ae721e36c12e0ea24fd23e2fed61e. Follow the advisory at https://github.com/openclaw/openclaw/security/advisories/GHSA-v3qc-wrwx-j3pw. If you cannot upgrade immediately, the actionable compensating control is to prevent agents from mutating exec-approval settings: restrict or disable the gateway config.patch and config.apply actions for agent-facing tool calls, and treat tools.exec.ask and tools.exec.security (plus the legacy tools.bash alias) as read-only in your deployment; the trade-off is that legitimate agent-driven config changes will also be blocked. Additionally, keep tools.exec.security on 'allowlist' rather than 'full' and monitor for config mutations touching those keys, accepting that this adds operational friction to any workflow that legitimately reconfigures execution policy.
Auth bypass in OpenClaw voice-call extension before 2026.2.1. EPSS 0.68%. PoC and patch available.
Privilege escalation in OpenClaw (pre-2026.3.28) allows unauthenticated remote attackers to gain administrative access b
OpenClaw versions 2026.2.22 through 2026.2.24 contain a privilege escalation vulnerability that allows authenticated att
An authorization mismatch vulnerability in OpenClaw versions prior to 2026.3.1 allows authenticated users with operator.
OpenClaw versions prior to 2026.1.29 automatically establish WebSocket connections to attacker-controlled gateway URLs e
Path traversal in OpenClaw through version 2026.3.23 enables unauthenticated remote attackers to read arbitrary files in
OpenClaw sandbox browser functionality launches x11vnc for noVNC observer sessions without requiring authentication, all
OpenClaw versions before 2026.2.26 allow authenticated attackers to write arbitrary files outside the workspace director
OpenClaw versions prior to 2026.2.22 contain a shell environment variable injection vulnerability in the system.run func
OpenClaw versions prior to 2026.2.22 contain a resource exhaustion vulnerability where the application fails to consiste
OpenClaw versions prior to 2026.3.1 contain a sandbox escape vulnerability that allows authenticated attackers with low
OpenClaw versions 2026.1.30 and below fail to validate Telegram webhook secret tokens when `channels.telegram.webhookSec
Same weakness CWE-862 – Missing Authorization
View allSame technique Authentication Bypass
View allShare
External POC / Exploit Code
Leaving vuln.today
EUVD-2026-25333
GHSA-7f25-vhj7-mxrc