inboundfy-validador-webinar-imagem
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as webinar copy, briefs, and visual assets, which is a potential surface for indirect prompt injection attacks.
- Ingestion points: Identified in
SKILL.mdas "Thumbnail, copy aprovada, brief e identidade visual". - Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the input data.
- Capability inventory: The skill primarily reads context files and generates reports. No high-risk capabilities such as network exfiltration, arbitrary command execution, or sensitive credential access were found.
- Sanitization: The skill does not mention explicit sanitization or validation logic for the content of the ingested text.
Audit Metadata