The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill focuses on providing best practices and operational workflows for a video clipping tool.
[INDIRECT_PROMPT_INJECTION]: The skill describes a workflow where the agent processes data from external sources (video transcripts and AI-generated clip candidates). While this introduces a theoretical surface for indirect prompt injection, the skill explicitly mandates human review for every output, which serves as a robust control against accidental or malicious instruction execution from processed data.

opus-clip