The Agent Skills Directory

[PROMPT_INJECTION]: The skill defines patterns for importing external JSON data which creates an indirect prompt injection attack surface. * Ingestion points: Untrusted data enters the context via file reading in importSave and string parsing in loadPrefab. * Boundary markers: No specific boundary markers or 'ignore' instructions are provided in the code examples. * Capability inventory: The skill's environment is restricted to the Read, Glob, and Grep tools. * Sanitization: The logic uses JSON.parse() for structural validation but does not perform content-level sanitization.
[EXTERNAL_DOWNLOADS]: The documentation references the pako library in a compression example. This is a well-known and standard library used for Zlib compression.

ecs-serialization