skills/egorfedorov/slot-casino-game-developer-skills-for-stake-engine/autoplay-system-designer/Gen Agent Trust Hub
autoplay-system-designer
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill includes a local validation script,
scripts/validate_autoplay_spec.py, which ensures that autoplay specifications adhere to a defined schema. The script performs local file operations and basic data validation without any network access or high-privilege operations.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through the validation of external JSON files, which could theoretically contain instructions intended to influence the agent.\n - Ingestion point:
scripts/validate_autoplay_spec.pyreads a file from a path provided as a command-line argument.\n - Boundary markers: The skill instructions do not explicitly define delimiters or specific 'ignore instructions' warnings for the content of the input specification.\n
- Capability inventory: The skill's primary capability is shell execution of its own validation script and subsequent processing of the result. It does not have broad network or system write permissions.\n
- Sanitization: The input is processed using standard JSON parsing and validated against mandatory keys (limits, stopConditions, ui, etc.), which significantly restricts the surface for malicious payload delivery.
Audit Metadata