duck-adventure

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill operates as a self-contained game engine. It performs no network requests and limits file system interaction to a dedicated state directory.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied game commands, presenting a minimal indirect injection surface.
  • Ingestion points: User commands processed during the turn loop (SKILL.md).
  • Boundary markers: The skill explicitly mentions redaction of secrets before state writes and uses defined game-end semantics.
  • Capability inventory: File writes are restricted to .duck-adventure/state.json and state.md via hooks/adventure.py. No network or shell execution capabilities exist in the engine script.
  • Sanitization: Commands are validated against a specific verb surface defined in references/PARSER.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 05:07 AM
Security Audit — agent-trust-hub — duck-adventure