Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessObtain sandboxed session access
-
DeliveryCall cross-agent sessions_spawn operation
-
ExploitCreate child process on unsandboxed agent
-
ExecutionSet sandbox.mode to off
-
ImpactExecute code outside confinement
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V13 - Configuration Permission assignment review
|
| Exploitation Conditions | Requires OpenClaw versions prior to 2026.3.1 with cross-agent sessions_spawn operations enabled. Attacker must have authenticated low-privilege sandboxed session access to spawn child processes with sandbox.mode disabled, bypassing runtime confinement. |
| Impact | The CVSS 3.1 vector (AV:N/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H) indicates network-accessible exploitation with high attack complexity requiring low privileges, leading to high impact across confidentiality, integrity, and availability. The 'AC:H' (high complexity) and 'PR:L' (low privileges required) suggest exploitation requires specific conditions and authenticated access, somewhat limiting the attack surface. The vulnerability is tagged as 'Authentication Bypass' by VulnCheck, indicating it undermines security controls. Without EPSS data or KEV listing, current exploitation likelihood appears theoretical rather than observed in the wild. However, the public disclosure via GitHub Security Advisory GHSA-p7gr-f84w-hqg5 significantly increases risk as technical details are available. This represents a genuine priority for organizations running OpenClaw in multi-tenant or security-sensitive environments where sandbox escape could lead to lateral movement or privilege escalation, but may be lower priority for isolated deployments with strong network segmentation. |
| Exploit ScenarioDerived | An authenticated attacker with low-level credentials on an OpenClaw deployment initiates a sandboxed session and leverages the cross-agent sessions_spawn functionality to create a child process. By exploiting the failure to inherit sandbox constraints, the attacker spawns a new runtime with sandbox.mode explicitly set to off, escaping the confined environment and gaining full system access to execute arbitrary code, access sensitive data, or pivot to other systems within the network. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Mar 21, 2026
OpenClaw versions prior to 2026.3.1 contain a sandbox escape vulnerability that allows authenticated attackers with low privileges to bypass runtime confinement restrictions. Attackers can exploit a flaw in cross-agent sessions_spawn operations to create child processes under unsandboxed agents, effectively disabling sandbox protections by setting sandbox.mode to off. While the CVSS score is 7.5 (High), there is no evidence of active exploitation (not in CISA KEV), though the vulnerability has been publicly disclosed through GitHub Security Advisories and VulnCheck, increasing the likelihood of proof-of-concept development.
RemediationAI
Upgrade OpenClaw to version 2026.3.1 or later, which addresses the sandbox inheritance enforcement flaw in cross-agent sessions_spawn operations. The patch details and upgrade instructions are available in the official GitHub Security Advisory at https://github.com/openclaw/openclaw/security/advisories/GHSA-p7gr-f84w-hqg5. Until patching is completed, implement defense-in-depth controls including restricting authenticated user privileges to trusted principals only, monitoring for unusual cross-agent spawn operations, implementing network segmentation to limit the blast radius of potential sandbox escapes, and enforcing strict authentication policies to reduce the pool of PR:L attackers. Review audit logs for any suspicious sessions_spawn activity that may indicate exploitation attempts.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in npm
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
OpenClaw is affected as identified by CPE cpe:2.3:a:openclaw:openclaw:*:*:*:*:*:*:*:*. The vulnerability is rooted in CWE-732 (Incorrect Permission Assignment for Critical Resource), where the application fails to properly enforce sandbox inheritance policies during cross-agent process spawning operations. When a sandboxed session initiates a sessions_spawn operation across different agents, the sandbox configuration is not correctly propagated to child processes, allowing them to run with sandbox.mode disabled. This breaks the fundamental security boundary that runtime sandboxing is designed to enforce, permitting confined processes to escape their restricted execution environment.
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
AV:N/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H
Primary rating from NVD · only source for this CVE.
CVSS VectorNVD
Lifecycle Timeline
3DescriptionCVE.org
OpenClaw versions prior to 2026.3.1 fail to enforce sandbox inheritance during cross-agent sessions_spawn operations, allowing sandboxed sessions to create child processes under unsandboxed agents. An attacker with a sandboxed session can exploit this to spawn child runtimes with sandbox.mode set to off, bypassing runtime confinement restrictions.
Executive ViewAI
OpenClaw versions before 2026.3.1 contain a high-severity sandbox escape vulnerability that allows authenticated users with low privileges to disable security restrictions and execute unrestricted code. This poses significant risk to organizations using affected versions, particularly in multi-tenant or high-security environments where sandbox isolation is a critical control.
Restrict authentication access to OpenClaw to high-trust users only; implement network-level segmentation between agent nodes to limit lateral movement; disable or restrict cross-agent session spawning functionality if business operations permit; enable comprehensive logging and alerting on sandbox.mode configuration changes and child process creation from sandboxed agents; monitor for unauthorized agent communication patterns; enforce principle of least privilege for all OpenClaw user accounts.
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
Attack Technique
Authentication bypass attacks exploit flaws in the verification mechanisms that control access to systems and applications.
Related Vulnerabilities in OpenClaw Multiple Authorization and Command Execution
View all 20Auth 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
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
Authentication bypass in OpenClaw before 2026.5.3 allows remote attackers with low privileges to receive agent responses
Directory-context bypass in OpenClaw before 2026.8.1 allows authenticated attackers with user interaction to reuse an 'a
EUVD-2026-13943
GHSA-wr92-6w3g-2hwc
NVD
CVE.org
EPSS
Vulnrichment
VulDB
Exploit-DB