safe-change

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides comprehensive security guardrails for AI-driven code development. It specifically instructs the agent to prevent the leakage of secrets, maintain authentication boundaries, and ensure data integrity.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided code change requests, establishing a standard interaction surface for indirect prompt injection.\n
  • Ingestion points: User-supplied requirements for code implementation as referenced in SKILL.md and agents/openai.yaml.\n
  • Boundary markers: The instructions do not define specific delimiters or isolation markers for the user-provided data.\n
  • Capability inventory: The agent is directed to perform repository analysis, modify source code, and execute verification tools like build systems or test runners as outlined in references/risk-guide.md.\n
  • Sanitization: The skill includes a mitigation strategy requiring the agent to surface material risks and decisions to the user for confirmation before proceeding.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 03:13 AM
Security Audit — agent-trust-hub — safe-change