inboundfy-base-imagem
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes untrusted data from the project context to influence its output.
- Ingestion points: The skill reads multiple external files including
context/marca-voz.md,context/canais.md,context/glossario.md, and visual identity definition files (palette, typography, logo, and grid) to guide the image generation process. - Boundary markers: There are no explicit instructions to use delimiters or protective wrappers when interpolating the content of these files into the agent's active prompt, which could allow instructions embedded in those files to influence agent behavior.
- Capability inventory: The skill has the capability to call external image generation tools available in the environment and write files to specific project directories (
97-ativos-finais/<canal>/<item>/). - Sanitization: The instructions do not specify any validation or sanitization routines for the content ingested from the Markdown or configuration files, relying on the presence of these files as a prerequisite for execution.
Audit Metadata