There's no equivalent mandate for software engineers. Nothing stops you from spending years as a pure "prompt pilot" and losing the ability to read a stack trace or reason about algorithmic complexity. The atrophy is silent and gradual.

The author's suggestion to write code by hand as an educational exercise is right but will be ignored by most, because the feedback loop for skill atrophy is so delayed. You won't notice you've lost the skill until you're debugging something the agent made a mess of, under pressure, with no fallback.

The deeper issue: most of these guardrails assume the threat is accidental (agent goes off the rails) rather than adversarial (something in the agent's context is actively trying to manipulate it). Time-boxed domain whitelists help with the latter but the audit loop at session end is still reactive.

The /revert snapshot idea is underrated though. Reversibility should be the first constraint, not an afterthought.

I'm sharing a new feature that I've added to NLUX: React Server Component Adapter

If you’re working with Next.js, building conversation AI, and love tinkering with generative UI, this might be a game-changer for you.

Ref short demo on X

Feel free to share your feedback or ask questions. Your insights are invaluable!

Happy coding!

Comments URL: https://news.ycombinator.com/item?id=40799816

Points: 1

# Comments: 1

On memory: You're in luck, we do support conversation history and we're actively working on improvements like lazy loading for long conversations.

Best practices: Spot on! The project docs have tips for building assistants that hide the underlying LLM ( Feature: Assistant personas - https://docs.nlkit.com/nlux/examples/assistant-persona ). Plus, docs cover LangChain integration (supports any AI backend!), Next.js, HuggingFace (access to thousands of models at your fingertips), and any custom backend.

I wanted to share a UI toolkit project I’ve been working on recently, born out of how difficult I found it to build a great UX on top of LLMs, and keep application state in sync. I’ve built:

- A React/JS front-end library for conversational interfaces, which makes it super easy to bootstrap AI assistants and ChatGPT style UX: https://github.com/nlkitai/nlux

- A set of adapters that simplify integration with AI backends such as LangServe and HuggingFace

The library is highly configurable, easy to theme, supports markdown streaming (that was tough to get right!), custom rendering for AI response components, and it works smoothly with Next.js, React JS (of course), and even plain JavaScript (for the Vue.js folks out there ;).

I built it with an architecture I’d love myself as a dev: high code modularity, zero dependencies, wide unit test coverage, and lots of documentation and examples.

You can play with codesandbox previews and demos here: https://docs.nlkit.com/nlux

Thanks for reading. It feels like we’re at a new frontier building with LLMs, and I’d love to hear what challenges you’ve had on integrating with them & building great UI & UX.

Comments URL: https://news.ycombinator.com/item?id=40658605

Points: 24

# Comments: 3

The real challenge would be to do it for their flagship products .. Basecmap, and more ..

Rather than hastily installing the latest shiny package, we should refactor and build reusable minimal utilities tailored to our domain.

Comments URL: https://news.ycombinator.com/item?id=39177269

Points: 1

# Comments: 0

I hope that you'll find it useful! :)

Comments URL: https://news.ycombinator.com/item?id=39113893

Points: 9

# Comments: 1