agent-loop
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to its ingestion of external project documentation. \n
- Ingestion points:
SKILL.mdreadsdocs/agents/loops.mdandCONTEXT.mdto configure the loop's logic and retrieve verification commands. \n - Boundary markers: No explicit delimiters or instructions are used to distinguish the skill's core logic from the content ingested from the project files. \n
- Capability inventory: The agent is empowered to modify files ('Act' phase) and execute shell commands ('Observe' and 'Verify' phases). \n
- Sanitization: The skill lacks mechanisms to sanitize or validate the verification commands it retrieves from documentation before executing them. \n- [COMMAND_EXECUTION]: The skill dynamically executes commands as specified in
docs/agents/loops.md. This allows external project documentation to dictate the shell commands run by the agent during the verification phase.
Audit Metadata