synthesia-avatar-video
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines workflows for processing untrusted user data, such as names, company details, and script text, to populate API requests for video generation. This creates an attack surface where malicious input could attempt to influence the generated content or bypass platform safety filters.
- Ingestion points: Untrusted data enters via
templateData(e.g.,first_name,account_name,cta_url) and thescriptTextfields inSKILL.md. - Boundary markers: The instructions recommend using
test: truefor watermarked previews and mandate a comprehensive human 'Safety gate' and 'QA' process before final production rendering. - Capability inventory: The skill utilizes network operations via
POSTandGETrequests to the official Synthesia API (api.synthesia.io). - Sanitization: The documentation explicitly instructs agents to perform preflight validation for 'forbidden claims', 'policy-sensitive language', and 'compliance review' to mitigate injection risks.
Audit Metadata