skills/egorfedorov/slot-casino-game-developer-skills-for-stake-engine/slot-mechanics-designer/Gen Agent Trust Hub
slot-mechanics-designer
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from JSON files to generate implementation specs and patch plans, creating an attack surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the context through files processed by scripts/check_mechanics_spec.py.
- Boundary markers: There are no specific instructions or delimiters used to prevent the agent from following instructions hidden within the specification data.
- Capability inventory: The skill allows file reading and generates command-line verification steps.
- Sanitization: Validation is limited to structural and logical integrity of the game mechanics, with no sanitization of natural language content.
Audit Metadata