arquitecto-de-briefing-landing

Pass

Audited by Gen Agent Trust Hub on Apr 12, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructions specify analyzing content from external platforms including Instagram and Google Maps/Business. This involves network-based retrieval of information from these well-known services to extract business details for project planning.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external sources and interpolates it directly into a prompt template for a subsequent agent.
  • Ingestion points: Data extracted from Instagram links, Google Maps/Business links, and free-form user text as described in the 'Ingesta y Extracción' section of SKILL.md.
  • Boundary markers: Not present. The skill does not define specific delimiters or instructions to prevent extracted content from being interpreted as commands by the AI.
  • Capability inventory: The generated output instructs a subsequent agent to prepare repositories on GitHub and call several other skills ('arquitecto-de-estructura-digital', 'curador-de-estilo-visual', 'auditor-estatico-plus'), creating a trust chain for potentially injected instructions.
  • Sanitization: Not present. There are no mechanisms mentioned for validating or escaping the content retrieved from external links before it is processed.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 12, 2026, 05:52 AM
Security Audit — agent-trust-hub — arquitecto-de-briefing-landing