vcs-commit-msg

Pass

Audited by Gen Agent Trust Hub on May 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local version control commands (git log, svn log --xml, hg, jj) to retrieve repository history for style analysis. This is a core functional requirement of the skill and is performed locally.
  • [PROMPT_INJECTION]: The skill is subject to an indirect prompt injection surface because it processes untrusted data from the repository's commit history (e.g., commit subjects and bodies authored by others).
  • Ingestion points: Project commit history retrieved via git log, svn log, etc.
  • Boundary markers: The skill specifies explicit delimiters (--- commit ---) to separate individual log entries.
  • Capability inventory: Access to shell execution for VCS tools and the ability to propose content to the commit buffer.
  • Sanitization: The skill relies on statistical analysis (80% agreement threshold) to determine style, which naturally mitigates the influence of isolated malicious commits.
  • [SAFE]: The skill implements a strong security posture by requiring the agent to show the final composed message to the user and wait for explicit approval before executing the final commit command.
Audit Metadata
Risk Level
SAFE
Analyzed
May 1, 2026, 07:43 PM
Security Audit — agent-trust-hub — vcs-commit-msg