godot-save-load-systems
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements logic for ingesting external data from save files in the persistent user directory, which represents a potential attack surface for indirect prompt injection if the files are manipulated by an untrustworthy source.
- Ingestion points: File reading operations in
scripts/save_load_patterns.gdandscripts/save_migration_manager.gdthat load data from theuser://filesystem. - Boundary markers: The skill utilizes structured data formats (JSON and binary Variants), which inherently provide field boundaries, though it does not specify LLM-specific delimiters.
- Capability inventory: The skill possesses the capability to perform filesystem read/write operations and asynchronous resource loading based on data extracted from these files.
- Sanitization: The instructions explicitly include hardening measures such as using
data.get()with default values, performing range checks on numeric inputs, and disabling object decoding (allow_objects=false) during deserialization to prevent RCE.
Audit Metadata