godot-analyst
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted Godot project data, including file structures, GDScript source code, and scene files, which creates an attack surface for indirect prompt injection.
- Ingestion points: The agent performs high-fidelity mapping of the
res://directory tree and audits individual script files (.gd) and scene files (.tscn). - Capability inventory: The skill uses GDScript helpers (
scoring_logic.gd,visionary_comparison.gd) to perform RegEx-based auditing and persists result baselines to the local filesystem (user://). It generates architectural critiques and certificates based on this data. - Sanitization: The skill does not explicitly define sanitization or escaping protocols for content extracted from user project files before it is processed by the agent's reasoning engine.
- Boundary markers: While the skill provides strict 'Operating Rules' in
SKILL.mdto guide agent behavior (e.g., mandatory rubric loading), it lacks explicit delimiters to distinguish untrusted project content from instruction context.
Audit Metadata