hydra

Local multi-model prompt lab — fan one prompt out across parallel model lanes with streamed output, token and cost tracking, persisted sessions, and a synthesis pass. Next.js, Prisma, Vercel AI SDK.