video-template-frame-glitch-title
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources and generative models via the SandBase API, creating a potential surface for indirect instructions to enter the agent's context.
- Ingestion points: Data and media retrieved from
sandbase_runas described inreferences/sandbase-api-map.md, and user-provided source material defined in theSKILL.mdworkflow. - Boundary markers: The skill includes a dedicated 'SandBase boundary' section that explicitly limits external tool usage to non-deterministic tasks and requires the agent to maintain local work boundaries.
- Capability inventory: The skill has access to
sandbase_discover,sandbase_inspect,sandbase_run,sandbase_run_get,sandbase_runs, andsandbase_accountfor interacting with the vendor's API. - Sanitization: The workflow includes a mandatory quality gate (Workflow Step 4) requiring verification of output through tests, previews, and source comparison rather than assuming success from file creation.
Audit Metadata