configure-physics-collision

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructions involve the processing of external data such as project settings, save files, and network snapshots which could serve as vectors for indirect prompt injection.
  • Ingestion points: External project settings, saved game states, and network snapshots (referenced in the 'Security Guardrails' section).
  • Boundary markers: None explicitly defined in the provided instructions for separating instructions from data.
  • Capability inventory: The skill is granted access to high-privilege tools including execute_command, write_file, edit_file, and run_tests.
  • Sanitization: The skill recommends validating deserialized physics state for finite values and sane magnitudes, and pinning engine versions to maintain determinism.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 11:20 AM