OpenClaw CLI versions from 2026.1.29 up to but not including 2026.6.6 have a vulnerability in the Browser CDP discovery feature that allows acceptance of blocked WebSocket URLs. This can enable lower-trust callers or configured input paths to reach network destinations that should be blocked by OpenClaw policy. The vulnerability is mitigated by upgrading to version 2026.6.6 or later, restricting the feature to trusted operators, or disabling it when not needed.
OpenClaw CLI versions from 2026.1.29 up to but not including 2026.6.5 have a vulnerability where MCP SSE redirects could forward Authorization headers. This may allow a lower-trust caller or input path to execute or persist actions beyond their intended authorization. The impact depends on the operator's configuration and whether untrusted input can reach the affected feature. The issue is fixed in version 2026.6.5. Users should upgrade to this version or restrict/disable the affected feature until patched.
OpenClaw CLI versions from 2026.1.29 up to but not including 2026.5.28 have a vulnerability where a workspace .env file can override the Homebrew executable selection during skill install flows. This allows a trusted operator opening a repository with a crafted .env file to cause an unintended Homebrew-compatible executable to run. The vulnerability is scoped to this feature and does not affect OpenClaw's trusted-operator model outside this context. The issue is patched in version 2026.5.28.
CVE-2026-53818 is a vulnerability in openclaw-cli where the MCP loopback feature could allow non-owner callers to bypass owner-only tool policies and before-tool-call hooks. This could lead to unauthorized invocation of owner-only behaviors if the feature is enabled and accessible. The impact depends on the operator's configuration and whether untrusted inputs can reach the vulnerable loopback path. A fix is available in version 2026.4.24. Until patched, restricting MCP loopback access to trusted operators and narrowing allowlists is recommended.
CVE-2026-53817 is a vulnerability in OpenClaw's Control UI affecting LAN/shared-token deployments where locality information is spoofed during pairing. An attacker with network or authentication access to the Control UI pairing path can obtain a durable admin device token. This token persists beyond shared token rotations unless explicitly removed. The flaw arises from overly trusting locality-derived signals for pairing decisions. The issue is fixed in OpenClaw version 2026.5.28.
OpenClaw CLI versions from 2026.1.29 up to but not including 2026.5.19 contain a vulnerability where message read actions can bypass channel allowlist checks. This allows lower-trust callers with access to these actions to read messages from channels they should not have access to, depending on operator configuration. The issue is fixed in version 2026.5.19.
OpenClaw versions from 2026.1.29 up to but not including 2026.5.28 have a vulnerability where hook-triggered CLI runs could gain owner-scoped MCP tool authority improperly. This occurs when hooks are enabled and a bundled CLI backend is selected, allowing a caller with a hook token to execute owner-only MCP tools. The issue is fixed in version 2026.5.28. The vulnerability has a high severity with a CVSS score of 8.4. Mitigation includes upgrading to version 2026.5.28 or later, keeping hook tokens secret, restricting network access to hook endpoints, and disabling hooks if not needed.
CVE-2026-53813 is a vulnerability in openclaw-cli where fake package roots can influence memory-core artifact loading. This occurs when a local package root resolution path, influenced by workspace state, selects an unintended package root. The vulnerability affects versions from 2026.1.29 up to but not including 2026.4.25. The impact depends on the operator's configuration and whether lower-trust input can reach the affected path. A fix is available in version 2026.4.25. Until patched, it is recommended to run memory-core flows from trusted workspaces and apply configuration hardening.
A vulnerability in OpenClaw's Matrix allowFrom feature allows binding to mutable display names, enabling a Matrix account that can change display name metadata to match a policy entry improperly. This could lead to unauthorized agent access intended for another Matrix identity if the feature is enabled and reachable. The issue is fixed in version 2026.5.7. Until patched, using stable Matrix user IDs in allowlists and restricting Gateway sharing between untrusted users are recommended mitigations.
OpenClaw CLI versions from 2026.1.29 up to but not including 2026.5.18 contain a vulnerability where marketplace runtime extension metadata can redirect runtime loading to unscanned payloads. This allows a trusted operator's selected package to load plugin code outside the reviewed package entry points. The vulnerability depends on the operator's configuration and the reachability of the affected feature. A fix is available in version 2026.5.18. Until patched, it is recommended to install only trusted plugins, maintain explicit plugin allowlists, and disable the affected feature if not needed.
OpenClaw CLI versions from 2026.1.29 up to but not including 2026.4.25 contain a vulnerability where embedded runner policy can be confused by provider aliases. This may cause policy checks to be performed against an alias rather than the canonical provider identity, potentially allowing access to bundled tools outside the intended provider policy. The vulnerability requires the affected feature to be enabled and reachable, and its practical impact depends on operator configuration and input trust boundaries. A fix is available in version 2026.4.25.
CVE-2026-53806 is a vulnerability in openclaw-cli where combined POSIX shell options can cause inconsistent parsing between approval-time and execution-time, potentially allowing inline shell commands to run without proper allowlist checks. This affects versions from 2026.1.29 up to but not including 2026.5.12. The vulnerability requires the affected feature to be enabled and reachable, and its impact depends on operator configuration and input trust levels. A fix is available in version 2026.5.12. Until patched, avoiding combined shell option forms in allowlisted commands is recommended.
OpenClaw's QQBot channel had a vulnerability where native approval buttons did not enforce the configured approver identity, allowing unauthorized QQ users to approve exec or plugin actions. This flaw affected versions from 2026.1.29 up to but not including 2026.5.18. The issue was specific to native approval buttons and did not affect text command approvals. A fix is available in version 2026.5.18.
OpenClaw CLI's Slack plugin approvals used the exec approver gate for plugin actions, allowing Slack users authorized only for exec approvals to resolve plugin approvals improperly. This issue affects versions from 2026.1.29 up to but not including 2026.5.12. The vulnerability depends on the operator's configuration and whether lower-trust input can reach the approval path. A fix is available in version 2026.5.12. Until patched, manual review of Slack approval actions and maintaining narrow allowlists are recommended mitigations.
OpenClaw versions from 2026.1.29 up to but not including 2026.5.18 contain a vulnerability where a paired node can forge exec lifecycle events without proper provenance verification. This allows a malicious or compromised paired node to send crafted event data to the gateway, causing it to treat attacker-supplied events as legitimate exec lifecycle results. The flaw requires the attacker to already control a paired node connection and does not allow unauthenticated access to the gateway. The vulnerability is addressed in OpenClaw version 2026.5.18.
OpenClaw CLI versions from 2026.1.29 up to but not including 2026.5.18 contain a vulnerability where certain browser control interactions can bypass private-network navigation checks. This allows an authenticated browser-control user to trigger navigation to private or loopback URLs via UI actions that evade initial SSRF protections, potentially exposing private page content to browser evaluation. The vulnerability does not allow unauthenticated access to OpenClaw and only affects environments where browser control is enabled and accessible to authenticated users. A fix is available in version 2026.5.18.
A path-confinement bypass vulnerability in openclaw-cli versions up to 2026.3.1 allowed writes outside intended root directories during browser output handling. This issue was fixed in version 2026.3.2 by unifying root-bound, file-descriptor-verified write semantics and canonical path-boundary validation across browser output and related install/skills write paths.
OpenClaw CLI versions from 2026.1.29 up to but not including 2026.2.14 are affected by a cross-site request forgery (CSRF) vulnerability in browser-facing localhost mutation endpoints. These endpoints accept cross-origin requests without proper Origin or Referer validation, allowing malicious websites to trigger unauthorized state changes in the victim's local OpenClaw browser control plane. The vulnerability is mitigated by rejecting mutating HTTP methods when requests indicate a non-loopback Origin or cross-site fetch context. A patch is available that enforces these checks, and enabling authentication on the browser control interface also mitigates risk.
OpenClaw CLI versions up to 2026.2.13 have a vulnerability in the Telnyx webhook handler of the optional voice-call plugin that allows unauthenticated HTTP POST requests if the Telnyx public key is not configured. This enables attackers to forge Telnyx events. The issue affects deployments where the voice-call plugin is enabled and the webhook endpoint is accessible to attackers. A fix is planned for version 2026.2.14 and later, which enforces signature verification by requiring the Telnyx public key configuration.