Set up Cloud Agent dev environment for Grok Forge

[xre217]

Summary

Sets up the Cursor Cloud Agent development environment for Grok Forge (Next.js 16 local AI studio, Ollama-powered) and documents non-obvious startup caveats for future agents.

The only code change is an added ## Cursor Cloud specific instructions section in AGENTS.md. Dependency installation is handled by the environment update script (npm install); Ollama and the llama3.2:3b model are baked into the VM snapshot (system dependencies, not part of the update script).

What was set up

• npm install — project dependencies (Next.js 16, React 19, AI SDK, etc.)
• Ollama installed (+ zstd, required by the installer) and llama3.2:3b pulled (~2 GB)
• .env.local created from .env.example (local-only mode, no xAI key)

Verification

Task	Command	Result
Lint	npm run lint	Runs; 14 pre-existing react-hooks errors in committed source (not env issues)
Build	npm run build	Compiles successfully (15 routes)
Run	npm run dev:local	Ready at http://localhost:3000/studio
Health	npm run verify	All 7 route checks pass
Hello-world	POST /api/chat + browser chat	Assistant reply from ollama · llama3.2:3b end-to-end

The current Ollama build runs the AMX ggml CPU backend fine on this VM — no backend workaround was needed.

Demo

Sent a chat message in the studio; the assistant replied locally via Ollama with an OLLAMA · LLAMA3.2:3B engine badge.

grok_forge_chat_hello_world.mp4

Grok Forge chat reply with Ollama badge

_{To show artifacts inline, enable in settings.}