The open source project named fulling, and it's hit 1k stars (github.com)

Fulling, an open-source project that recently hit 1,000 stars, packages a turnkey, AI-first full‑stack sandbox that wires Claude Code, Next.js, and a dedicated PostgreSQL instance into a single developer experience. In about a minute it provisions a Next.js (App Router) + shadcn/ui TypeScript app, a KubeBlocks‑backed PostgreSQL database, the Claude Code CLI, HTTPS ingress with auto‑mapped subdomains, and a ttyd web terminal that accepts natural‑language instructions so an AI agent can write, run, and deploy code. The repo uses Prisma, NextAuth (GitHub OAuth), Tailwind v4, a custom fullstack-web-runtime image, and Kubernetes for orchestrating isolated per‑project sandboxes with secrets, network policies, resource limits, and WebSocket support for terminals. For the AI/ML community this matters because Fulling drastically reduces setup friction for AI‑driven full‑stack development and experimentation: environment variables (API keys, DATABASE_URL, AUTH config) are auto‑injected, projects can sync to GitHub, and sandboxes can be promoted to high‑availability production clusters. Key implications include faster prototyping of agent‑authored applications, reproducible infrastructure for research, and built‑in security/isolation primitives; caveats are requirements for a Kubernetes cluster with KubeBlocks, cloud credentials, and secure secret management when exposing AI agents to live infra.