itgoyo-ip-builder
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of user-provided images and character descriptions to generate IP packages.
- Ingestion points: Processes external character photos and textual role descriptions as specified in
SKILL.mdandreferences/turnaround-workflow.md. - Boundary markers: The skill uses structured templates for character specifications and requires explicit user confirmation of features before finalization.
- Capability inventory: The skill possesses file system write capabilities to create directories and save character assets.
- Sanitization: Implements input normalization (kebab-case for IDs) and explicitly forbids the use of absolute paths or parent directory references (
..) to prevent path traversal attacks during file writing. - [COMMAND_EXECUTION]: The skill manages files and directories within a designated path (
~/.agents/itgoyo-ip-builder). It includes specific logic to prevent overwriting existing data by automatically incrementing version numbers and ensures that operations remain within intended directory boundaries.
Audit Metadata