design-system-adr
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/adr.pyusessubprocess.runto invoke thegitandgh(GitHub CLI) binaries. These calls are used to determine the repository's root directory and to fetch ADR numbers currently claimed in open pull requests. The implementation uses list-based argument passing rather than shell execution, which correctly mitigates potential command injection vulnerabilities. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and lint Architecture Decision Records (markdown files) which may contain content from multiple contributors. While the tool itself only performs regex-based metadata extraction and structural validation, the agent reading these files could be susceptible to instructions embedded in the decision text. The ingestion point is the
documentation/adr/directory, and while the tool ensures specific sections are present, it does not sanitize or provide boundary markers for the actual content of the decision bodies.
Audit Metadata