aos-fatos
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill operates within a strictly defined scope, limited to a single verified domain (aosfatos.org). All link generation and content processing are designed to prevent hallucination and unauthorized data access.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the web, which is a common surface for indirect prompt injection. This is mitigated by several layers of protection: 1. Ingestion points: Data enters the agent context through WebFetch results in the 'Passo Fetch' section of SKILL.md. 2. Boundary markers: The skill instructions include an explicit 'Guardrail contra prompt injection' that commands the agent to ignore instructions found within fetched content. 3. Capability inventory: The skill uses a local script (scripts/aosfatos_helper.py) for URL construction and deduplication, and the agent's native fetch tool for network requests. 4. Sanitization: The ValidadorDeItens class in the helper script ensures all processed URLs belong to the valid domain (aosfatos.org) before they are returned to the user.
Audit Metadata