spine-asset-prep

Pass

Audited by Gen Agent Trust Hub on Nov 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted layered image projects (e.g., PSD, .pcraft) and JSON manifest files containing asset metadata. Maliciously crafted layer names or metadata within these files could attempt to influence agent behavior during ingestion.
  • Ingestion points: Reads project layer trees, image dimensions, and manifest.json files for handoff data.
  • Boundary markers: No explicit delimiters or "ignore embedded instructions" warnings are provided for the data processing stages.
  • Capability inventory: Employs functions.exec for tool interaction and manipulates files via MCP interfaces, providing significant influence over the local environment.
  • Sanitization: Mentions geometric and hash-based validation (SHA-256), but lacks specific sanitization for text-based metadata fields that could host injection payloads.
  • [COMMAND_EXECUTION]: The instructions guide the agent to use functions.exec to discover and invoke functionality within external MCP tools (PhotoCraft, Photoshop). This capability allows the agent to execute specific commands on the system through these tool interfaces.
  • [DYNAMIC_EXECUTION]: The skill specifies that "local code" may be used for pixel reading, contour analysis, and geometry calculations. This indicates the agent may generate or execute scripts at runtime to perform advanced image processing tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 15, 2026, 01:02 PM