refine-review

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or security vulnerabilities were detected. The skill's behavior is consistent with its stated purpose of artifact review and quality assurance.
  • [DATA_EXFILTRATION]: The skill accesses local project files such as PBIs and documentation within the team-artifacts/pbis/ and .claude/skills/shared/ directories. There are no network operations (e.g., curl, wget, fetch) or patterns suggesting the exfiltration of this data to external services.
  • [COMMAND_EXECUTION]: No shell commands or subprocess calls were identified. The skill relies on standard platform tools like TaskCreate and AskUserQuestion for its operations and task management.
  • [PROMPT_INJECTION]: The skill uses authoritative language (e.g., 'MANDATORY IMPORTANT MUST') to define its operational logic and ensure adherence to review protocols. This is a standard instructional technique for defining agent behavior and does not represent an attempt to bypass agent safety controls or override system instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 07:25 AM
Security Audit — agent-trust-hub — refine-review