debugging
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform system-level diagnostics, including git history analysis (git log), working tree status checks (git status), and file searches (find, grep). These operations are restricted to read-only forensic analysis and do not modify the environment.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, specifically git commit messages and plan files, which constitutes a vulnerability surface. Ingestion points: Git history and task plan files (SKILL.md, Phase 1). Boundary markers: Not explicitly defined for data interpolation. Capability inventory: Subprocess calls via Bash for git and filesystem operations across ALL scripts. Sanitization: The skill implements a mandatory 'Scrub' step (SKILL.md, Phase 3) that explicitly redacts potential secrets, including API keys (sk-), tokens (ghp_), and passwords, as well as absolute home paths from the final report before it is presented to the user.
Audit Metadata