introspection
Pass
Audited by Gen Agent Trust Hub on May 12, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill defines a set of high-level principles and checklists for an AI agent to monitor its own performance, quality, and error patterns. It contains no instructions for bypassing safety filters or exfiltrating data.
- [COMMAND_EXECUTION]: The skill provides instructions on how to use the 'llmem' CLI tool. This includes commands such as
llmem introspect,llmem search, andllmem add. These commands are used for managing local memory and logs relevant to the agent's own tasks. - [DATA_EXPOSURE]: The skill references local session files located in
~/.local/share/opencode/sessions/. This path is used by the agent to analyze its own history of actions and is not an attempt to harvest sensitive user credentials like SSH keys or cloud provider configurations.
Audit Metadata