worldview-gameplay-experience-study
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a pure instruction-based tool designed for game design research. It operates entirely on text templates and local markdown/SVG reference files provided within the skill package.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external gameplay evidence (timecoded video analysis or descriptions). While this represents an attack surface for processing untrusted data, the skill includes strict boundary instructions to de-identify source data and avoid copying distinctive assets. Furthermore, the skill lacks dangerous capabilities such as network access, file system write operations, or command execution, limiting the impact of potential injection.
- Ingestion points: Gameplay evidence provided in the user's initial prompt (SKILL.md, README.md).
- Boundary markers: Instructions explicitly separate observed evidence from reconstructed experience and mandate the use of de-identified study IDs (SKILL.md).
- Capability inventory: Limited to text generation; no subprocess calls, file writes, or network operations detected across scripts or instructions.
- Sanitization: Instructions require stripping source URLs, recognizable game names, and distinctive dialogue from the output.
Audit Metadata