godot
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines logic for processing external asset packs, including metadata from JSON files and Tiled maps, which creates a vulnerability surface for instructions embedded in data.\n
- Ingestion points: The skill (SKILL.md) specifies reading
atlas JSON,styles.json,display.json, and.tmj(Tiled) files to drive project configuration and asset mapping.\n - Boundary markers: While the skill mentions the pack should be "validated" by a prior process, it does not instruct the agent to use specific boundary markers or to ignore instructions embedded within these data files.\n
- Capability inventory: The skill performs file system organization (copying folders to
res://), resource life-cycle management (queue_free()), and resolution settings based on the ingested data.\n - Sanitization: No explicit sanitization or strict schema validation is described for the ingested JSON or TMJ content within this skill's scope.
Audit Metadata