stress-test

Pass

Audited by Gen Agent Trust Hub on Aug 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill constructs image generation prompts by interpolating asset descriptors and reference file paths directly from project documentation (e.g., asset passports, registry files). This creates a surface for indirect prompt injection where malicious instructions embedded in these files could influence the agent's interaction with the image generation model.\n
  • Ingestion points: Data is ingested from CLAUDE.md, AGENTS.md, /docs/registry.md, and asset passport files located in /assets/characters/, /assets/locations/, and /assets/props/.\n
  • Boundary markers: The skill lacks explicit boundary markers or instructions to ignore embedded commands, instead requiring that descriptors be pasted verbatim.\n
  • Capability inventory: The skill has read/write access to the local project structure and interacts with an external image generation workflow.\n
  • Sanitization: No sanitization, escaping, or validation of the ingested content is performed before it is included in the output prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 16, 2026, 02:31 PM
Security Audit — agent-trust-hub — stress-test