inboundfy-arquitetura-site
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external inputs such as project objectives, content IDs, and persona definitions to generate site architecture maps, which creates a potential surface for indirect prompt injection.
- Ingestion points: Described in SKILL.md under 'Entrada esperada', where the agent receives external objectives and IDs.
- Boundary markers: The skill references a shared 'preflight' and 'contract' system (e.g.,
_shared/01-preflight-e-fontes.md) to establish context and artifact integrity. - Capability inventory: The skill reads local project configuration files (e.g.,
.inboundfy/context/empresa.md) and generates structured markdown documents with site maps and internal links. - Sanitization: The instructions explicitly require the use of validation skills like
inboundfy-anti-slopandinboundfy-copy-edicaoto review and edit content before delivery.
Audit Metadata