Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessOperator hot-disables memory mid-turn
-
ExploitStale enabled config retained by resolver
-
ExecutionExisting memory tool instance stays authorized
-
ImpactModel reads durable memory post-revocation
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V8 - Authorization Authorization logic validation
MITRE ATT&CK
1 technique across 1 tactic: Privilege Escalation The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Requires that OpenClaw memory features be in use and that an operator hot-disables the memory configuration WHILE a previously created memory_search/memory_get tool instance remains active within an already-running agent turn. The stale access only persists for the remainder of that single run, and it is read-only over durable memory (no writes, no availability impact). Limiting factors: high attack complexity, a specific mid-turn disablement timing window, low-privilege access already present, and passive user interaction - the flaw cannot be triggered against a fresh turn or after the run ends. This is not remotely exploitable against a default configuration by an unauthenticated attacker. |
| Impact | This is a genuinely low-priority issue despite being network-reachable on paper. The vendor CVSS 4.0 vector (AV:N/AC:H/AT:P/PR:L/UI:P/VC:L/VI:N/VA:N) scores just 2.1, and all signals agree it is minor: no EPSS score published, not on CISA KEV, and no public exploit identified. The high attack complexity (AC:H), attack requirement (AT:P), low-privilege requirement (PR:L) and passive user interaction (UI:P) reflect that this is not attacker-initiated in the traditional sense - it is a race-like revocation gap that only manifests during an already-running agent turn where a memory tool instance was created before the operator hot-disabled memory. Confidentiality impact is Low and integrity/availability are None, since the flaw only lets the model continue read-only access to durable memory for the remainder of the current run. There is no conflict between the description and CVSS: both point to a transient, low-impact authorization-enforcement lapse, not a widely weaponizable vulnerability. Real-world priority is low; patch on normal cadence. |
| Exploit ScenarioDerived | An operator hot-disables memory configuration while an agent turn is already in progress and a memory_search/memory_get tool instance was created before that change. Because the execution-time resolver restores the stale enabled configuration, the model keeps reading durable memory for the rest of that run despite the operator having revoked access. No public exploit code exists, and exploitation is opportunistic rather than attacker-driven, requiring the precise timing of disablement mid-turn. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Sep 26, 2026
OpenClaw versions before 2026.8.1 continue to honor memory_search and memory_get tool calls after an operator hot-disables the memory configuration, letting the model inside an already-running agent turn keep reading durable memory for the remainder of that run. Exploitation is authenticated (low privilege, PR:L) and depends on a narrow timing window in which memory is disabled while a previously created memory tool instance is still active, plus passive user interaction; the impact is limited to read-only disclosure of memory content with no writes or availability effect. There is no confirmed active exploitation and no public exploit code identified at time of analysis, and the issue is rated low severity (CVSS 2.1 / assessed CVSS 3.1 vector AV:N/AC:H/PR:L/UI:R/S:U/C:L/I:N/A:N) - a genuinely low-priority issue despite being network-reachable on paper.
RemediationAI
Vendor-released patch: upgrade the openclaw npm package to 2026.8.1 or later, which is the version the vendor identifies as fixing this memory access-control bypass; see the GitHub security advisory at https://github.com/openclaw/openclaw/security/advisories/GHSA-chr6-w4m5-57fv and the VulnCheck advisory at https://www.vulncheck.com/advisories/openclaw-before-2026.8.1-memory-access-control-bypass for details. Because the stale authority persists only for the remainder of an already-running agent turn, an effective compensating control when immediate upgrade is not possible is to terminate or forcibly restart any in-flight agent runs after disabling memory configuration, so no tool instance created under the enabled configuration survives the change - the trade-off is loss of in-progress turn state and interruption of user sessions. Organizations can additionally gate the memory_search and memory_get tools at an external policy or tool-invocation layer that re-checks current configuration on every call rather than trusting the creation-time snapshot, though this requires custom wrapper code and must be maintained alongside upstream releases. As an interim containment measure, restricting who can reach the agent endpoint (authenticated low-privilege users are sufficient to attempt this, and user interaction is required) reduces the pool of potential triggers, and monitoring for memory tool invocations issued after a configuration change can flag the window; neither measure eliminates the flaw, so upgrading to 2026.8.1 remains the primary fix.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
OpenClaw is a Node.js agent runtime distributed as the npm package 'openclaw', in which tools such as memory_search and memory_get are instantiated with a configuration snapshot captured at creation time. The execution-time resolver is intended to re-evaluate configuration on each invocation, but it treats an explicit disablement the same way it treats an unavailable configuration snapshot and falls back to restoring previously captured (stale) authority. This is a CWE-863 Incorrect Authorization defect: an authorization decision made at tool-creation time is never invalidated when the operator revokes the underlying capability mid-run. Because the stale instance lives for the duration of the current agent turn, the model can continue querying durable memory after the operator disabled it, although the flaw cannot be triggered against a freshly created turn and does not persist after the run ends. The affected component is the memory tool layer of the agent framework, not the underlying memory store, and the exposure is read-only. The CPE cpe:2.3:a:openclaw:openclaw:*:*:*:*:*:*:*:* confirms all package versions prior to the fix are in scope.
Know when Openclaw gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:H/AT:P/PR:L/UI:P/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N
AC:H and UI:R reflect the narrow mid-turn revocation-timing window; PR:L for an already-active low-privilege session; C:L read-only memory, no integrity or availability impact.
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
Lifecycle Timeline
1DescriptionCVE.org
OpenClaw (npm package 'openclaw') before 2026.8.1 fails to revoke memory tool access when an operator hot-disables memory configuration. Existing memory_search and memory_get tool instances retain the enabled configuration captured at creation time because the execution-time resolver treats explicit disablement like an unavailable configuration snapshot and restores the stale authority. As a result, during an already-running agent turn the model can continue searching and reading durable memory after the operator revoked that access, for the remainder of that run. Exploitation requires memory to be disabled while a previously created memory tool remains active. The issue is fixed in 2026.8.1.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
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
Directory-context bypass in OpenClaw before 2026.8.1 allows authenticated attackers with user interaction to reuse an 'a
Unauthenticated access to the noVNC observer interface in OpenClaw's Docker sandbox browser allows anyone who can reach
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
EUVD-2026-87511
GHSA-9r83-7xq3-vjrp
NVD
CVE.org
EPSS
Vulnrichment
Exploit-DB