soul-audit
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the shell tool to invoke the gbrain command-line utility for managing identity state and rendering files. Specific commands include gbrain bootstrap interview and gbrain bootstrap render.
- [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by processing untrusted user input that influences agent identity and behavior.
- Ingestion points: User answers are captured via gbrain bootstrap interview --set and stored in state/interview.json (SKILL.md).
- Boundary markers: The instructions do not specify the use of delimiters or boundary markers for the user-provided identity content.
- Capability inventory: The skill possesses the capability to execute shell commands and modify critical system files such as SOUL.md, USER.md, ACCESS_POLICY.md, and HEARTBEAT.md (SKILL.md).
- Sanitization: There is no mention of sanitization or filtering of user-provided strings before they are rendered into identity prose.
Audit Metadata