migration

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious behavior, obfuscation, or unauthorized access patterns were detected. The skill uses standard placeholders and references internal project documentation appropriately.- [PROMPT_INJECTION]: The skill handles untrusted user input via the $ARGUMENTS placeholder within <migration-description> tags, identifying an indirect prompt injection surface. This is mitigated by the explicit requirement to present designs and obtain user consent before any filesystem changes.
  • Ingestion points: User input is ingested into SKILL.md via the $ARGUMENTS variable.
  • Boundary markers: The input is enclosed in <migration-description> XML-style tags.
  • Capability inventory: The skill is designed to generate migration files, implying file-write capabilities.
  • Sanitization: There is no programmatic sanitization; however, the skill mandates a human-in-the-loop approval step before file operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 07:25 AM
Security Audit — agent-trust-hub — migration