21st-video
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill performs legitimate network operations to the vendor's infrastructure, such as uploading media assets to a dynamically provided URL and exporting rendered videos. These operations are required for the skill's primary functionality.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface by reading untrusted data from the current project.
- Ingestion points: Processes the project's README, UI components, brand assets, and release notes to establish video content (SKILL.md).
- Boundary markers: No explicit delimiters or instructions are provided to the model to ignore potentially malicious embedded instructions in the ingested files.
- Capability inventory: The skill uses a suite of
video_*tools to modify project state and export files. - Sanitization: The instructions recommend inspecting validation details and schemas returned by the tool, which acts as a technical constraint on the injected content.
- [CREDENTIALS_UNSAFE]: The skill manages temporary "handoff tokens" in editor URLs. However, the instructions explicitly mitigate the risk of credential exposure by directing the agent never to print the handoff token or persist the URL to logs or repositories.
Audit Metadata