semantic-git

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill is susceptible to indirect prompt injection because it identifies and executes CI scripts from untrusted local files.\n
  • Ingestion points: Reads script names and execution patterns from package.json, justfile, makefile, and pyproject.toml (as specified in references/ci-verification.md).\n
  • Boundary markers: Absent; the skill does not implement delimiters or 'ignore' instructions for data read from the repository context.\n
  • Capability inventory: Possesses shell command execution capabilities via git, npm, pnpm, make, cargo, and pytest.\n
  • Sanitization: Absent; the skill trusts and executes scripts found in the repository configuration without validation.\n- [Command Execution] (LOW): The skill's primary function is the execution of shell commands.\n
  • Evidence: The skill constructs and executes git commands and various CI tools. Although it includes a 'User-In-The-Loop' protocol for commit operations, CI checks are executed automatically upon detection of relevant configuration files.\n- [Dynamic Execution] (LOW): Constructs shell commands dynamically at runtime using local file data and AI-generated messages.\n
  • Evidence: Assembles complex git commit strings incorporating environment variables like GIT_AUTHOR_DATE and GIT_COMMITTER_DATE based on repository history and user preferences.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:15 PM