specialist-engineering-git-workflow-master

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill interpolates user input directly into the system instructions for a sub-agent without proper encapsulation or boundary markers. \n
  • Ingestion points: The user's question is directly inserted into the Agent tool prompt via the {the user's question} placeholder in SKILL.md. \n
  • Boundary markers: Absent. The prompt does not use delimiters or instructions to ignore potential commands embedded in the user input. \n
  • Capability inventory: The sub-agent is defined as a Git strategy specialist; the skill does not explicitly provide tools for file-system writes, code execution, or network access beyond the Agent delegation tool itself. \n
  • Sanitization: Absent. No filtering or validation of the user-supplied string is performed before interpolation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 08:50 PM
Security Audit — agent-trust-hub — specialist-engineering-git-workflow-master