test-audit
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a variety of local binaries located in
~/.claude/skills/gstack/bin/(such asgstack-skill-start,gstack-paths,gstack-slug, andgstack-skill-end) along with standard command-line tools likegit,grep,xargs, andawkto perform repository analysis and audit tasks.- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external data including repository files, code comments, and git history to identify candidate tests for removal. While this exposes a surface for indirect prompt injection, it is addressed by instructions to treat repository content as data rather than instructions and by using theAskUserQuestiontool to require human confirmation for destructive actions.- [SAFE]: The skill includes robust security directives, such as prohibiting the dumping of environment variables or full process tables to prevent credential exposure. It also enforces a 'User-origin gate' that validates configuration tuning requests against the user's chat history to prevent profile poisoning from external files.
Audit Metadata