godot-rea
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from a
ledger.jsonfile to populate Godot project configurations, scene nodes, and GDScript stubs. This ingestion of external data into the agent's workspace creates a potential attack surface for indirect prompt injection. - Ingestion points: Fields such as
id,name,detail, andevidencefrom theledger.jsonfile entries. - Boundary markers: The skill uses single-line comments in generated GDScript files to separate ledger-derived metadata from executable code.
- Capability inventory: The script has the ability to create directories, write text files (scripts, scenes, config), and perform binary file copies.
- Sanitization: The skill implements defensive measures including regex-based identifier cleaning (
[A-Za-z0-9_]+) and whitespace normalization in comments (stripping newlines) to prevent breakout from the intended structure. - [DYNAMIC_EXECUTION]: The skill generates GDScript (
.gd) and Godot scene (.tscn) files at runtime. While the generated scripts are stubs using simple templates, the process involves assembling executable content from external data sources.
Audit Metadata