This sends video files to Gemini's flash model and returns structured breakdowns of what's actually in the frame: shot beats, voiceover, on-screen text, pacing, and creative strategy. Built for analyzing TikTok or Reels at the video level, not just guessing from metadata or transcripts. It handles local MP4s and URLs, keeps source paths so you can link results back to your dataset, and writes natural Markdown reports instead of raw JSON. The scope is practical short-form structure, things like hook timing and product placement. If you're running creative QA or building a catalog, this gives you observable evidence to work from instead of surface-level stats.
npx -y skills add postplusai/postplus-skills --skill video-analysis --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills