kf-git-commit-japanese-commit-message
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill contains no instructions for network access, sensitive file manipulation, or privilege escalation.
- [NO_CODE]: The skill consists exclusively of markdown instructions and does not include any scripts or executable code.
- [PROMPT_INJECTION]: The skill processes untrusted user input (descriptions of code changes) to generate formatted commit messages, creating an indirect prompt injection surface. This is assessed as safe due to the complete lack of exploitable tool capabilities. 1. Ingestion points: User-provided descriptions of changes or existing commit messages (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: None; the skill does not use subprocesses, file-writing tools, or network operations. 4. Sanitization: Absent.
Audit Metadata