smart-git-automation
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill constructs and executes shell commands ("git checkout", "git add", "git push", "gh pr create") using dynamically generated branch names and file paths. Although the instructions caution against unsafe string concatenation for filenames, the execution of commands based on file system data creates a potential injection surface.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local data to generate commit messages and PR descriptions, creating an attack surface for indirect instructions embedded in repository content.
- Ingestion points: "git status" and "git diff" outputs are used to analyze changes and group files (SKILL.md).
- Boundary markers: The skill lacks specific delimiters or instructions to ignore embedded content within the analyzed diffs.
- Capability inventory: The agent has permissions to write to the file system via git operations and interact with network services via "git push" and "gh pr create" (SKILL.md).
- Sanitization: While the skill advises on safe pathspec handling to prevent shell injection, it lacks sanitization for the generated text derived from untrusted file content, which could be exploited to influence the agent's logic or downstream CI/CD systems.
Audit Metadata