ssh-doctor

Warn

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies heavily on sudo to execute system-level commands, including launchctl for service management, systemsetup for Remote Login configuration, and kill for terminating processes.
  • [CREDENTIALS_UNSAFE]: There is a dedicated section for managing OP_SERVICE_ACCOUNT_TOKEN which involves scanning shell profiles for 1Password Service Account tokens.
  • [DATA_EXFILTRATION]: The skill includes a routine to retrieve sensitive tokens from a remote host via SSH (ssh ... HOST 'awk ...') and save the output locally, effectively automating the movement of secrets across the network.
  • [COMMAND_EXECUTION]: The skill modifies the user's ~/.profile by appending exported environment variables and management markers, which is a common persistence mechanism used to alter shell behavior.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 13, 2026, 10:53 AM
Security Audit — agent-trust-hub — ssh-doctor