ketchup-video
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists entirely of natural language instructions for the agent. It does not provide any scripts, binaries, or configuration files that execute logic automatically.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data (screenshots, diffs, recordings, and project data) to generate visual summaries. This represents a potential surface for indirect prompt injection if the ingested data contains malicious instructions meant to influence the video content or agent behavior. However, this is a standard risk for any summarization tool and is mitigated by the agent's internal safety guardrails.
- [COMMAND_EXECUTION]: The instructions permit the agent to write and render video files (e.g., mp4) outside the current repository. While this is an exception to a "read-only" rule, it is necessary for the skill's primary function and does not involve arbitrary or dangerous system commands.
Audit Metadata