Takes an existing Veo clip and adds more frames after it. You pass a source video URL plus a prompt describing what happens next, and Veo 3-1 continues the clip with consistent motion, lighting, and subject identity. Two endpoints: full quality for final delivery, fast for iteration. Useful when you hit the 8-second generation cap and need 16 seconds total, or when you're chaining narrative beats shot by shot from a single seed. The docs warn that identity drift stacks up after three or four chained extends, so keep individual extensions short if you're building a longer sequence. Routes through the runcomfy CLI with proper exit codes and JSON schema validation.
npx skills add https://github.com/doany-ai/skills --skill video-extendjuliusbrussee/caveman
mattpocock/skills
mertbuilds/skills
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills