OpenClaw Security Crisis 2026: What Happened and Lessons Learned

Security stories about an agent spread quickly because the software can act on a computer. A useful postmortem still needs a source, a version boundary, and a clear statement of what was tested. This page is a checklist for reading those reports, not a reconstruction of an incident that this publication cannot independently verify.

Start with the primary record

Use the OpenClaw Security Advisories, the versioned releases, and the Gateway security documentation. A secondary report can point you to a lead; it cannot replace the affected-version range or remediation steps in the primary record.

What operators can do now

• Run openclaw security audit after installation and after changing exposure.
• Keep direct messages behind pairing and groups behind an allowlist or mention gate.
• Review skills, plugins, MCP servers, and provider credentials before enabling them.
• Use a separate Gateway or host when users do not share a trust boundary.
• Preserve logs and the exact version when investigating a suspicious action.

openclaw security audit
openclaw doctor

These controls reduce exposure; they do not make an agent risk-free. Follow the official release notes for patches and migrations, and report a suspected vulnerability through the repository's responsible disclosure process.