material-maker

Pass

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes shell commands via the Bash tool to interact with the Material Maker CLI for exporting textures and to run the local scripts/preflight.py utility for file integrity checks.
  • [DYNAMIC_EXECUTION]: The documentation identifies that Unreal Engine 5 (UE5) export workflows generate Python scripts at runtime. The skill explicitly instructs the agent to require manual user review and approval of these generated scripts before any execution occurs.
  • [DATA_EXFILTRATION]: To mitigate risks of accidental data exposure, the skill includes a preflight.py script that contains a blacklist of sensitive file paths (e.g., .ssh, .aws, .env) and keywords (e.g., secret, token, credential) used to prevent the agent from validating or accessing private data.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external files like .ptex node graphs and 3D models. Ingestion points include project loading and CLI exports. Boundary markers are established by requiring explicit user approval for remote data and generated code. Capabilities include file system access and shell execution. Sanitization is achieved through pre-execution hashing of known files and manual artifact review.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 6, 2026, 08:52 AM