design
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
SKILL.mdfile documents installation procedures for external modular skills using a local Python script (scripts/install-external-skills.py). This is a standard project-management operation within the agent's expected workflow. - [INDIRECT_PROMPT_INJECTION]: The
scripts/generate_design_brief.pyutility processes external markdown content (Product Requirement Documents) to generate documentation skeletons. - Ingestion points: The script reads markdown files from paths provided as command-line arguments.
- Boundary markers: None; the script relies on regex patterns to identify markdown headers and does not process instructions within the content.
- Capability inventory: The script's operations are limited to reading local files and writing text to the standard output.
- Sanitization: No content sanitization is performed, as the script is intended for structural extraction of text for human review.
- [DATA_EXPOSURE]: The
hooks/pre-tool.shscript scans the project directory for filenames matching patterns like 'brief' or 'prd' to provide advisory warnings if design documentation is missing. This search is confined to the local project scope and is used for process compliance.
Audit Metadata