1
I've been trying out different monitor layouts that are more optimal to work with AI Agents. I put a top monitor that works almost like a HUD (Head-Up Display), with all the agent CLIs running.
My main monitor will be my own work area, but the top one is just a bunch of agents running, so I can monitor them easily by just looking up. If I need to interrupt or work with the agents, I can just look up. I found this much better than using side monitors for posture.
What's your setup?
I've been working on n.codes, an open-source framework that lets end users generate their own UI within your app using natural language prompts.
The problem: Every SaaS team drowns in feature requests. Most aren't complicated—just different arrangements of existing data: "show me refunds over $500 that need approval," "export this report formatted for accounting," "bulk archive these records."
The backend can already do it. The bottleneck is exposing capabilities through UI.
How it works:
- Reads your components (React/Vue) to understand available UI primitives and styles
- Builds a capability map from your APIs, schemas, and docs
- When a user prompts, generate a sandboxed UI in your app that uses existing APIs and components
It's not for marketing pages or complex flows. It's for the 80% of internal tools that are just CRUD, dashboards, and workflows.
Current status: Early stage. The architecture is defined, core capability mapping works. Looking for feedback before building out the full runtime.
Looking for feedback on:
- Does this solve a real problem you've experienced?
- What would make you trust generated UI in a production app? (sandbox model, permissions, audit trails?)
- Similar approaches I should study?
GitHub: https://github.com/yungookim/n.codes Site: https://n.codes