git-commit-composer
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it reads and processes unvalidated content from git diff outputs. This is an inherent property of any tool that analyzes source code changes and is not indicative of malicious design. \n
- Ingestion points: The skill executes
git diff --cachedandgit diffto ingest staged and unstaged code changes (SKILL.md). \n - Boundary markers: None identified; the diff output is directly analyzed by the agent without specific delimiters. \n
- Capability inventory: The skill is designed only to read diffs and produce descriptive text; it does not have the capability to execute code or access the network based on the diff content. \n
- Sanitization: No sanitization is applied to the diff output as the full context is required for accurate commit message generation. \n- [COMMAND_EXECUTION]: The skill runs the
gitcommand to analyze the repository state. This is a fixed, safe use of a standard CLI tool for its intended purpose and does not involve dynamically constructing commands from external input.
Audit Metadata