skills/karlorz/agent-skills/autopilot/Gen Agent Trust Hub

autopilot

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements an 'autopilot' loop using standard Claude Code hooks. It creates repo-local configuration in .claude/ to manage task continuity, which is the stated and intended purpose of the tool.
  • [COMMAND_EXECUTION]: The installer script and hook templates utilize standard file system operations and shell commands (e.g., chmod, cat, jq) to manage session state and permissions. These operations are restricted to the local project and temporary system directories.
  • [SAFE]: The skill incorporates loop protection by implementing a configurable turn limit (defaulting to 20 turns) and providing a dedicated command (/autopilot_reset) to monitor and stop the autonomous process.
  • [SAFE]: The /autopilot_reset command uses a benign dynamic context injection (!bash) to display session status. This operation is limited to reading status flags from the skill's own temporary files in /tmp.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 02:37 PM