cut-planner

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill logic is confined to processing local video project metadata. It does not attempt to access sensitive system files, perform network requests, or execute arbitrary code.
  • [PROMPT_INJECTION]: The skill ingests data from clip summaries, representing a low-risk attack surface for indirect prompt injection.
  • Ingestion points: Processes library.yaml and markdown summaries from the local directory.
  • Boundary markers: Absent; the agent reads files without explicit separation of data and instructions.
  • Capability inventory: Restricted to local file updates and triggering a subsequent planning skill.
  • Sanitization: No sanitization is performed on external content.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 09:47 AM