pm-brand-motion
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local shell script named
check-update.shvia a relative path (../../check-update.sh) during its preamble. This introduces a dependency on external local logic not contained within the primary skill file. - [PROMPT_INJECTION]: The skill processes content from external Markdown files (
原型设计方案.md,产品定位方案.md) to generate new design assets, creating a surface for indirect prompt injection. - Ingestion points: Reads project-specific brand and prototype data from files in the
docs/02-方案设计/directory. - Boundary markers: The skill does not implement delimiters or system instructions to ignore potential injections within the read files.
- Capability inventory: The skill utilizes shell execution, file reading, and file writing tools.
- Sanitization: Ingested content is interpolated into generated HTML and Markdown files without escaping or validation.
Audit Metadata