CVE-2026-53813

OpenClaw before 2026.4.25 contains a path traversal vulnerability in memory-core artifact loading where workspace state influences local package root resolution. Attackers with access to affected workspaces can load memory-core artifacts from unintended local locations, potentially executing...

Runtime Skill Audit: Targeted Runtime Probing for Agent Skill Security

Agent skills let LLM agents reuse instructions, resources, tools, and workflows, but they also create a new place for malicious behavior to hide. A skill may look benign in its documentation or code while becoming harmful only when it is invoked with particular user requests, local assets,...

Updating the taxonomy of failure modes in agentic AI systems: What a year of red teaming taught us

In this article 1. Why the Taxonomy Needed Updating 2. Seven new failure modes 3. Operational findings: What red teaming showed 4. New mitigations 5. What to do this quarter When the Microsoft AI Red Team published the Taxonomy of Failure Modes in Agentic AI Systems in April 2025, the goal was a...

EUVD-2026-28163

OpenClaw versions 2026.2.21 before 2026.4.10 contain an authentication bypass vulnerability in the sandbox noVNC helper route that exposes interactive browser session credentials. Attackers can access the noVNC helper route without bridge authentication to gain unauthorized access to the...

EUVD-2026-28192

OpenClaw before 2026.4.22 contains a time-of-check/time-of-use race condition in the OpenShell filesystem bridge that allows attackers to read files outside the intended mount root. Attackers can exploit symlink swaps during filesystem operations to bypass sandbox restrictions and access...

CVE-2026-43530

OpenClaw versions 2026.2.23 before 2026.4.12 contain a weakened exec approval binding vulnerability in busybox and toybox applet execution that allows attackers to obscure which applet would actually run. Attackers can exploit opaque multi-call binaries to bypass exec approval mechanisms and weak...

CVE-2026-42436 OpenClaw < 2026.4.14 - Internal Page Content Exposure via Browser Snapshot and Screenshot Routes

OpenClaw before 2026.4.14 contains an improper access control vulnerability in browser snapshot, screenshot, and tab routes that fail to consistently validate the final browser target after navigation. Authenticated callers can bypass SSRF restrictions to expose internal or disallowed page conten...

CVE-2026-41910

OpenClaw before 2026.4.8 omits owner-only enforcement for cross-channel allowlist writes in the /allowlist endpoint. An authorized non-owner sender can bypass access controls to perform allowlist modifications against different channels, violating the intended trust model...

EUVD-2026-26111

OpenClaw before 2026.3.31 contains an incomplete scope-clearing vulnerability in trusted-proxy authentication mode that allows operator.admin privilege escalation. Attackers can exploit this by declaring operator scopes on non-Control-UI clients, allowing self-declared scopes to persist on...

CVE-2026-41398

OpenClaw before 2026.4.2 contains an improper access control vulnerability in the iOS A2UI bridge that treats generic local-network pages as trusted origins. Attackers can inject unauthorized agent.request runs by loading attacker-controlled pages from local-network or tailnet hosts, polluting...

CVE-2026-41392

OpenClaw before 2026.3.31 contains an exec allowlist bypass vulnerability allowing attackers to inherit allowlist trust via shell init-file wrapper invocations. Attackers can exploit shell options like --rcfile, --init-file, and --startup-file to load attacker-chosen initialization files while...

OpenClaw 后置链接漏洞

OpenClaw is an open-source intelligent artificial assistant developed by OpenClaw. Versions of OpenClaw prior to 2026.3.31 had a post-link vulnerability due to a sandbox escape issue. This vulnerability could allow remote attackers to access arbitrary files by exploiting symbolic links during fil...

GHSA-C4QG-J8JG-42Q5 OpenClaw: QQBot direct media upload skipped URL SSRF validation

Affected Packages / Versions - Package: openclaw npm - Affected versions: 2026.4.20 - Patched version: 2026.4.20 Impact The QQBot direct-upload media path could forward attacker-controlled image URLs without applying the SSRF validation used by the local download path. This could make configured...

Incomplete List of Disallowed Inputs

Overview openclaw is a 🦞 OpenClaw — Personal AI Assistant Affected versions of this package are vulnerable to Incomplete List of Disallowed Inputs in the environment variable loading process. An attacker can influence trusted runtime behavior by setting specially crafted OPENCLAW variables in a...

CVE-2026-41299

OpenClaw before 2026.3.28 contains an authorization bypass vulnerability in the chat.send gateway method where ACP-only provenance fields are gated by self-declared client metadata from WebSocket handshake rather than verified authorization state. Authenticated operator clients can spoof ACP...

DNS Rebinding

Overview openclaw is a 🦞 OpenClaw — Personal AI Assistant Affected versions of this package are vulnerable to DNS Rebinding via improper hostname validation in the browser navigation policy. An attacker can access internal network resources or sensitive endpoints by exploiting DNS rebinding...

CVE-2026-35628

OpenClaw before 2026.3.25 contains a missing rate limiting vulnerability in Telegram webhook authentication that allows attackers to brute-force weak webhook secrets. The vulnerability enables repeated authentication guesses without throttling, permitting attackers to systematically guess webhook...

OpenClaw 安全漏洞

OpenClaw is an open-source intelligent artificial assistant developed by OpenClaw. Versions of OpenClaw prior to 2026.3.23 contained security vulnerabilities. These vulnerabilities stemmed from a replay vulnerability in Plivo V2 signature verification, allowing attackers to bypass replay protecti...

OpenClaw: iOS A2UI bridge trusted generic local-network pages for agent.request dispatch

Summary Before OpenClaw 2026.4.2, the iOS A2UI bridge treated generic local-network pages as trusted bridge origins. A page loaded from a local-network or tailnet host could trigger agent.request dispatch without the stricter trusted-canvas origin check. Impact A loaded attacker-controlled page...

Incorrect Authorization

Overview openclaw is a 🦞 OpenClaw — Personal AI Assistant Affected versions of this package are vulnerable to Incorrect Authorization in the process that fetches thread root and reply context, which bypasses the sender allowlist. An attacker can gain unauthorized access to message threads by...