background-process-start-process
Fail
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill defines a tool 'start_process' in 'SKILL.md' that accepts an arbitrary 'command' string. This provides a direct interface for executing any shell command on the underlying host system without validation or sanitization.
- [REMOTE_CODE_EXECUTION]: The ability to execute arbitrary shell commands can be leveraged to download and execute malicious payloads or scripts from remote servers using tools like curl or wget.
- [DATA_EXFILTRATION]: This command execution capability can be exploited to read sensitive system files, environment variables, or SSH keys and transmit them to external, attacker-controlled servers.
Recommendations
- AI detected serious security threats
Audit Metadata