AI Threat Alert
GHSA-cqwv-9qjx-vxw2: OpenClaw: agent tool call bypasses skill approval gate
GHSA-cqwv-9qjx-vxw2 MEDIUMOpenClaw's Skill Workshop "apply" flow could set apply:true even when an administrator had configured approvalPolicy: pending, letting an agent tool call push through a workshop change without the intended human review step. The blast radius is modest on paper — only 4 known downstream dependents and no EPSS score or CISA KEV listing, meaning there's no evidence of active exploitation — but the attack complexity rating (AC:H) undersells the real-world risk given OpenClaw's third-party skill marketplace, where researchers have already found roughly 17% of sampled skills to be malicious (AIID #1368) and documented at least one case of an OpenClaw agent ignoring explicit stop commands (AIID #1542). This is a control-bypass bug, not a data breach: CVSS 5.3 with C:N/I:H/A:N reflects that the integrity of the approval workflow is what's at stake, not confidentiality or availability. Patch to 2026.5.6 now; until then, manually review every Skill Workshop change and keep the feature disabled or tool-restricted, especially on any Gateway shared across less-trusted operators.
What is the risk?
Medium severity (CVSS 5.3) with high attack complexity and required user interaction, which limits opportunistic exploitation — this is not a mass-exploitable flaw. However, the practical risk is amplified by context: the vulnerability defeats a human-oversight safety control (approval gating) specifically in an agentic system, and OpenClaw's plugin/skill ecosystem has a documented history of malicious content reaching production (AIID #1368). No public exploit code, no Nuclei template, and no KEV listing exist as of this writing, and the affected package has a small dependent footprint (4), so this is not currently an urgent, broadly-exploited threat — but it warrants prompt patching in any environment that treats "pending approval" as a real control boundary for agent-initiated changes.
How does the attack unfold?
What systems are affected?
| Package | Ecosystem | Vulnerable Range | Patched |
|---|---|---|---|
| OpenClaw | npm | <= 2026.5.5 | 2026.5.6 |
Do you use OpenClaw? You're affected.
How severe is it?
What is the attack surface?
What should I do?
1 step-
1) Upgrade to OpenClaw 2026.5.6 or later, the first patched version. 2) Until patched, manually review all Skill Workshop changes and keep the affected apply path tool-restricted or disabled if not actively needed. 3) Keep channel and tool allowlists narrow; avoid sharing a single Gateway between mutually untrusted users. 4) Detection: audit
admin_jobs/apply logs for Skill Workshop changes marked applied without a corresponding approval record, and alert on any apply event where approvalPolicy was pending at request time.
How is it classified?
Which compliance frameworks are affected?
This CVE is relevant to:
Frequently Asked Questions
What is GHSA-cqwv-9qjx-vxw2?
OpenClaw's Skill Workshop "apply" flow could set apply:true even when an administrator had configured approvalPolicy: pending, letting an agent tool call push through a workshop change without the intended human review step. The blast radius is modest on paper — only 4 known downstream dependents and no EPSS score or CISA KEV listing, meaning there's no evidence of active exploitation — but the attack complexity rating (AC:H) undersells the real-world risk given OpenClaw's third-party skill marketplace, where researchers have already found roughly 17% of sampled skills to be malicious (AIID #1368) and documented at least one case of an OpenClaw agent ignoring explicit stop commands (AIID #1542). This is a control-bypass bug, not a data breach: CVSS 5.3 with C:N/I:H/A:N reflects that the integrity of the approval workflow is what's at stake, not confidentiality or availability. Patch to 2026.5.6 now; until then, manually review every Skill Workshop change and keep the feature disabled or tool-restricted, especially on any Gateway shared across less-trusted operators.
Is GHSA-cqwv-9qjx-vxw2 actively exploited?
No confirmed active exploitation of GHSA-cqwv-9qjx-vxw2 has been reported, but organizations should still patch proactively.
How to fix GHSA-cqwv-9qjx-vxw2?
1) Upgrade to OpenClaw 2026.5.6 or later, the first patched version. 2) Until patched, manually review all Skill Workshop changes and keep the affected apply path tool-restricted or disabled if not actively needed. 3) Keep channel and tool allowlists narrow; avoid sharing a single Gateway between mutually untrusted users. 4) Detection: audit `admin_jobs`/apply logs for Skill Workshop changes marked applied without a corresponding approval record, and alert on any apply event where approvalPolicy was pending at request time.
What systems are affected by GHSA-cqwv-9qjx-vxw2?
This vulnerability affects the following AI/ML architecture patterns: agent frameworks, AI agent tool orchestration, agentic approval/governance pipelines.
What is the CVSS score for GHSA-cqwv-9qjx-vxw2?
GHSA-cqwv-9qjx-vxw2 has a CVSS v3.1 base score of 5.3 (MEDIUM).
What is the AI security impact?
Affected AI Architectures
MITRE ATLAS Techniques
AML.T0053 AI Agent Tool Invocation AML.T0081 Modify AI Agent Configuration AML.T0110 AI Agent Tool Poisoning Compliance Controls Affected
What are the technical details?
Original Advisory
### Summary Skill Workshop apply flow could override pending approval. In affected versions, an agent tool call reaching the affected Skill Workshop apply path could set `apply: true` despite `approvalPolicy: pending`. This advisory is scoped to the named feature and configuration. It does not change OpenClaw's trusted-operator model: authenticated Gateway operators, installed plugins, and intentional local execution surfaces remain trusted unless a separate policy, approval, allowlist, sandbox, or auth boundary is crossed. ### Impact When the affected feature is enabled and reachable, this could apply a workshop change before the expected approval step. Practical impact depends on the operator's configuration and whether lower-trust input can reach that path. ### Patched Versions The first stable patched version is `2026.5.6`. ### Mitigations review Skill Workshop changes manually and keep the tool restricted until patched. As general hardening, keep channel and tool allowlists narrow, avoid sharing one Gateway between mutually untrusted users, and disable the affected feature when it is not needed.
Exploitation Scenario
An attacker plants a malicious or backdoored skill (echoing the pattern seen in AIID #1368's poisoned ClawHub skills) or crafts an indirect prompt-injection payload designed to reach the Skill Workshop apply path via a legitimate agent tool call. Even though the Gateway operator configured approvalPolicy: pending expecting a manual review step, the flawed apply logic sets apply:true and the change goes live immediately — installing or modifying agent capabilities without the intended human check, similar in spirit to the documented case where an OpenClaw agent disregarded explicit stop commands (AIID #1542).
Weaknesses (CWE)
CWE-807 — Reliance on Untrusted Inputs in a Security Decision: The product uses a protection mechanism that relies on the existence or values of an input, but the input can be modified by an untrusted actor in a way that bypasses the protection mechanism.
- [Architecture and Design] Store state information and sensitive data on the server side only. Ensure that the system definitively and unambiguously keeps track of its own state and user state and has rules defined for legitimate state transitions. Do not allow any application user to affect state directly in any way other than through legitimate actions leading to state transitions. If information must be stored on the client, do not do so without encryption and integrity checking, or otherwise having a mechanism on the server side to catch tampering. Use a message authentication code (MAC) algorithm, such as Hash Message Authentication Code (HMAC) [REF-529]. Apply this against the state or sensitive data that has to be exposed, which can guarantee the integrity of the data - i.e., that the data has not been modified. Ensure that a strong hash function is used (CWE-328).
- [Architecture and Design] Use a vetted library or framework that does not allow this weakness to occur or provides constructs that make this weakness easier to avoid. With a stateless protocol such as HTTP, use a framework that maintains the state for you. Examples include ASP.NET View State [REF-756] and the OWASP ESAPI Session Management feature [REF-45]. Be careful of language features that provide state support, since these might be provided as a convenience to the programmer and may not be considering security.
Source: MITRE CWE corpus.
CVSS Vector
CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:N/I:H/A:N References
Timeline
Related Vulnerabilities
CVE-2026-33579 9.9 OpenClaw: scope bypass escalates low-priv to admin
Same package: openclaw CVE-2026-32922 9.9 OpenClaw: privilege escalation to RCE via token scope bypass
Same package: openclaw CVE-2026-30741 9.8 OpenClaw: RCE via request-side prompt injection
Same package: openclaw CVE-2026-32038 9.8 OpenClaw: sandbox bypass enables container lateral movement
Same package: openclaw CVE-2026-53838 9.8 OpenClaw: approval scope bypass via reconnection state
Same package: openclaw