worldview-game-safe-room-pressure-reset
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it is instructed to ingest and act upon data from external sources that could be controlled by an attacker.
- Ingestion points: The skill is directed to "Read local repository instructions" and process the "project and refuge brief" provided during the
/worldview-game-safe-room-pressure-resetinvocation (SKILL.md). - Boundary markers: While the skill uses a structured "Lock" system and markdown templates to organize data, it lacks hardened delimiters or specific "ignore embedded instructions" warnings when processing raw content from project files.
- Capability inventory: The agent is authorized to "identify normal run/test commands" and "execute the matrix" (SKILL.md), which involves running shell commands and writing implementation files to the source tree.
- Sanitization: The instructions do not define sanitization or validation protocols for the external commands or repository instructions identified during the analysis phase.
Audit Metadata