wachi-definicion

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted product descriptions and previous file versions to create structured contracts and update the product state machine.
  • Ingestion points: The skill reads fichero_input and existing file content during Phase 1 to understand the product context.
  • Boundary markers: There are no explicit instructions to use delimiters or ignore potentially malicious content within these inputs.
  • Capability inventory: The skill utilizes guardar_ficha, actualizar_dor_ficha, and transition_ficha to persist data, update definitions of ready (DoR), and transition the product lifecycle status.
  • Sanitization: No validation or sanitization of the input data is mentioned before it is committed to the versioned product record.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 04:39 PM
Security Audit — agent-trust-hub — wachi-definicion