orchestrate-batch-refactor

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests untrusted code from external repositories to drive the refactoring logic and generate work packets.\n
  • Ingestion points: Target repository files and modules specified during the analysis phase (SKILL.md).\n
  • Boundary markers: Absent. The instructions and templates lack explicit delimiters or instructions to ignore malicious directives embedded within the processed code comments or documentation.\n
  • Capability inventory: The skill involves file-writing operations (implied by the refactoring goal) and shell command execution for validation checks (e.g., npm run test mentioned in work-packet-template.md).\n
  • Sanitization: Absent. There is no automated validation or escaping of repository content before it is processed by the Explorer and Worker sub-agents.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 03:11 PM