DesktopCommander and Taskmaster are great to start with. With just these, you may start to see why OP recommends a memory MCP too (I don’t have a recommendation for that yet)
Nice to see new things. Does your team know about SwarmUI[1]?
SwarmUI has been nice for sharing workflows with friends who might not want to learn ComfyUI, they can type in a prompt or any other config exposed in the workflow and even upload image input. It’s pretty good.
I would still like to see a web interface for workflows that is more so built for indie hackers. I’d like to see a Next.js frontend that calls ComfyUI on a server, it has a specific workflow configured for that app, and the server could be reused for another Next.js frontend, all that needs to change is the workflow. Letting you create multiple web apps with 1 Comfy backend.
Yes, I just used GPT-4 to create a prompt for GPT-3.5-Turbo based on some loose rules that I laid out. It helped me fill in the gaps and write it in a concise format.
The prompt gave much much better results than than the one I wrote.
I know there are other options, but here's the relevance:
"Go, also known as Golang, is recognized for its simplicity, efficiency, and cross-platform capabilities. Its ease of use has made it an attractive choice for malware authors seeking to create versatile and sophisticated threats."
They also mention it's a "growing trend in malware development"
For the foundational parts I'd start with HTML, JS, and CSS syntax. It's very approachable as a beginner. I'd learn to use the browser devtools, learn how to use the CLI effectively (especially npm and git to get started) and I'd probably start with React or Next.js to start building web apps asap. They'll be around for a while it seems and they're extremely popular. Unfortunately, that only covers the frontend.
Next, I'd start learning to build APIs. I'd probably start with Node.js even today if I were to start again (even though it's not my favorite today). I'd start with maybe a simple CRUD app and if Copilot is an option, I'd try to see how much I could get it to write. Copilot does really well with "business logic". I'd use JSON to send responses to the frontend for this API and by then, I'd understand what full stack web development is a little bit. I'd also want to use an ORM in the beginning. I wish I had known about ORMs for working with databases sooner.
From there, I'd either A. Try to build something I've always wanted or B. Learn about a new frontend/backend technology C. Look at what others are building on X, LinkedIn, or HN.
If I was really focused on building AI products, I'd learn about LangChain and some of the OpenAI APIs to start.
To make the most of LLMs/Copilot, here are a few things that've helped me:
- Name variables, functions, etc. appropriately. Add comments to be explicit about something.
- Use the most popular tools & frameworks. Copilot has more data on them.
- ChatGPT can be even more useful than Copilot sometimes. Sometimes it's worse.
- Instead of Googling for the CLI command, ask ChatGPT.
Try Cmd+h to quickly hide a window. Night Shift is also nice to have. Cmd+tab for switching between apps. Alt+backspace is really handy (it deletes the previous word.) In fact, the alt key has a lot of functions that are often overlooked.
Rectangle is awesome for macOS. It makes it easier to put windows in the corner of the screen, etc. I also like to install Dozer to keep the menu bar clean on macOS.
As a new Phoenix convert I agree with a lot of these points from my experience too. The functional components make more sense in the majority of cases (and I think the docs point this out) and having the duplicate auth logic with on_mount feels less than ideal.
Hopefully the team will see this and make it even better!
Personally, I find myself accepting that the first draft will always have its flaws. Even after all these years it can be hard for me to see these flaws until the first draft is done.
Once the first draft is finished, I can look back on what could be done better and I improve it. Sometimes it requires a rewrite, sometimes it doesn't.
If I was confident that the thing I'm building can be done with a no-code solution, I'll try it out. But often the things I'm building certainly won't have a good no-code solution.
Learn what's available, learn how others have built similar software, and then just try to build that thing!