The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes project-level context files that could potentially contain untrusted instructions.
- Ingestion points: The skill reads data from context/canais.md, context/pessoas.md, and context/marca-voz.md in SKILL.md.
- Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between data and potential instructions within these context files.
- Capability inventory: The skill has the ability to write image assets to 97-ativos-finais/ and invoke other skills such as inboundfy-base-imagem and inboundfy-validador-webinar-imagem.
- Sanitization: There is no evidence of content escaping or validation performed on the ingested context data before it is used to generate the image briefs.
[SAFE]: The skill does not perform any network operations, access sensitive system credentials (like .ssh or .aws), or execute remote code from untrusted sources.
[SAFE]: All referenced skills and tools appear to be internal to the 'Inboundfy' framework, representing standard vendor functionality without evidence of typosquatting or impersonation.

inboundfy-especialista-webinar-imagem