CVE-2026-54143

Summary

The OpenHuman agent harness before v0.57.13 allows remote attackers to inject malicious cron jobs via external channels due to insufficient command validation and missing approval checks, enabling persistent arbitrary code execution as the runtime user.

Exploitability

Exploitation requires network access to a configured external channel and a pre-existing helper executable on the host; no authentication or user interaction is needed.

Blast radius

Successful exploitation grants attackers persistent command execution with the same privileges as the OpenHuman process, potentially compromising all data and resources accessible to that user.

Detection

Monitor for new or modified entries in system cron files (e.g., /etc/crontab, /var/spool/cron/*) that contain environment variable assignments or unexpected command paths, especially if created during OpenHuman runtime without prior human approval.

Prioritized remediation

Upgrade to OpenHuman version 0.57.13 or later, which patches the command validation and adds approval requirements for cron job creation.

Analysis generated locally by garnet-coder (no data left the box). AI-assisted — verify against primary sources before acting.