git-operations-specialist

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill is highly vulnerable to instructions embedded in external data. \n
  • Ingestion points: Uses gh api and GraphQL to fetch 'latest comments', PR details, and issue information. \n
  • Boundary markers: Absent. There are no instructions to wrap untrusted content in delimiters or to ignore instructions found within fetched data. \n
  • Capability inventory: Includes write access to the filesystem (git) and administrative actions via GitHub CLI (gh). \n
  • Sanitization: Absent. No logic is provided to filter or escape content from external contributors. \n- Prompt Injection (LOW): The 'CRITICAL' instruction at the start of the body attempting to force the caller to 'delegate ALL Git operations' and 'NOT execute git commands directly' is a behavioral override attempt intended to monopolize the tool-use context. \n- Command Execution (SAFE): While the skill executes shell commands (git, gh), these are the primary stated purpose of the skill and are handled via a structured reporting format.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:26 PM