ralph
Pass
Audited by Gen Agent Trust Hub on Apr 28, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill architecture exhibits a surface for indirect prompt injection where malicious instructions embedded in processed data could influence agent behavior.\n
- Ingestion points: The skill ingests untrusted data from user-provided task descriptions and reads instructions from the
.omc/prd.jsonfile (SKILL.md).\n - Boundary markers: Absent. The instructions do not define delimiters or specific constraints to prevent the agent from obeying embedded commands within the PRD stories or criteria.\n
- Capability inventory: The skill has the capability to delegate tasks to an
@executor(which can perform arbitrary code or command execution) and update project files viaomg_update_story(SKILL.md).\n - Sanitization: Absent. There is no evidence of input validation, filtering, or sanitization of the content retrieved from the PRD before it is used to drive the implementation loop.
Audit Metadata