creator-commerce
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided marketing briefs and data, which is a potential surface for indirect prompt injection. The skill implements a rigorous 'Evidence Discipline' to mitigate this risk.
- Ingestion points: User prompts containing marketing signals, data, and context provided during the 'frame' and 'evidence' workflow steps in SKILL.md.
- Boundary markers: The agent is required to classify all material claims as 'fact', 'inference', 'assumption', or 'unknown' as defined in references/skill-spec.json and SKILL.md.
- Capability inventory: The skill utilizes local scripts (dynamic_router.py, neural_router.py) and hooks (creator-measurement-check.md, evidence-gate.md) for routing and validation.
- Sanitization: The evidence-gate hook and strict invariants in references/skill-spec.json explicitly prohibit the fabrication of evidence or tool outputs.
- [COMMAND_EXECUTION]: The skill incorporates calls to local Python scripts (../../scripts/dynamic_router.py and ../../scripts/neural_router.py) to manage complex routing and dependency graphs. These are internal project scripts used for orchestration and do not involve external code downloads.
Audit Metadata