personal-ip-builder
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill generates new instruction sets by writing SKILL.md and configuration files to a local .agents/skills/ directory. This behavior is the intended secondary function of the skill to provide automated project-specific illustrations.
- [COMMAND_EXECUTION]: The instructions suggest running validation scripts available in the environment after the creation of the sub-skill.
- Evidence: Found in Phase 6 of SKILL.md regarding running environment validation scripts.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-provided images and interview answers to influence the generation of character contracts and sub-skills.
- Ingestion points: Reference images in Phase 1 and design interview responses in Phase 2 of SKILL.md.
- Boundary markers: Requires an explicit user confirmation step for the 'Character Contract' before proceeding with automated generation.
- Capability inventory: File writing (markdown and images), image generation tool calls, and potential validation script execution.
- Sanitization: Relies on human-in-the-loop verification of the generated contract rather than automated sanitization of the input data.
Audit Metadata