video-thumbnail-generator
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates within its intended scope of providing thumbnail generation instructions based on local context and video metadata. It does not exhibit signs of prompt injection, command execution, or unauthorized network activity.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface for indirect prompt injection as it processes external, untrusted data such as video transcripts.
- Ingestion points: Step 2 of the workflow in
SKILL.mdinstructs the agent to read project states and video transcripts. - Boundary markers: The instructions do not include specific delimiters or directives to disregard potential instructions embedded within the transcript data.
- Capability inventory: The agent's capabilities are restricted to frame inspection and prompt generation for image creation, with no direct access to system commands or exfiltration tools.
- Sanitization: There is no mention of sanitizing or filtering the content of ingested transcripts before they are processed by the agent.
Audit Metadata