fable-simulator
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external code changes and specifications to derive verification oracles, which represents an indirect prompt injection surface.
- Ingestion points: The skill ingests verification targets, public interfaces, and specifications provided as context to construct independent models (SKILL.md, references/oracle-derivation-guide.md).
- Boundary markers: A hard boundary between simulated evidence and real runtime verification is required, and simulation evidence must be distinctly labeled (Stage 7, SKILL.md).
- Capability inventory: The skill protocol includes executing test suites (e.g., bun test), performing headless browser interactions, and managing sandbox state transitions (SKILL.md, templates/oracle-contract.template.md).
- Sanitization: The skill enforces isolation through temporary directories and sandboxes and explicitly forbids the modification or deletion of untracked user files (references/oracle-independence-and-simulation-boundaries.md).
Audit Metadata