NYC

here-be-git

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • COMMAND_EXECUTION (LOW): The skill executes git init and git commit via the system shell. These are standard repository management operations and are performed only after explicit user intent is established.
  • PROMPT_INJECTION (LOW): The skill modifies CLAUDE.md and AGENTS.md to establish persistent instructions for the agent's workflow. While this is a core feature, it represents a mechanism for altering future agent behavior across sessions.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill ingests user input to generate the .gitignore file, creating a surface for potential instruction embedding.
  • Ingestion points: User-provided flavors or patterns in Step 3 of SKILL.md.
  • Boundary markers: Absent. User input is written directly into the .gitignore file.
  • Capability inventory: Shell execution (git init, git commit) and file writing (AGENTS.md, CLAUDE.md, .gitignore) within SKILL.md.
  • Sanitization: Absent. No validation or filtering is applied to the user-provided patterns before they are committed to the file system.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 08:03 AM