@promptpls

“There are like fifty tools here and I just want to build one app. Where do I actually start without reading all of this?”

“So one tool, one small project, come back when I'm stuck.”

“I only ever used the gemini command sometimes. Do I have to do anything by June 18?”

“And if 20 a day isn't enough?”

“This is a lot of steps. Do I really need all of them just to build something?”

“So step two — build the smallest version — is really where I start.”

“The terminal genuinely scares me. Is Claude Code actually usable if I've never really used a command line?”

“So once it's installed, it's really just chatting plus an approve button.”

“I'm building a small app. Does the model honestly matter, or is this just nerd stuff? They all write code.”

“So pick the best one for when it gets hard, not when it's easy.”

“If Clerk handles my users, do I still need a database? I'm confused about what goes where.”

“So Clerk for 'who is this person,' my database for 'what have they got.'”

“I'm brand new and just want one to install tonight. Which?”

“I just use the gemini command sometimes to ask questions in my terminal. Do I actually have to do anything, or will it keep working?”

“Smaller how much?”

“I'm new and just want autocomplete that doesn't cost money. Do I install Cursor or Cline for that?”

“So what do I use Cline and Cursor's local mode for then?”

“Honest question — if I move off the builder, am I now on my own with a pile of code I don't understand?”

“So it's the same way of working, just with the guardrails off.”

“I already use Claude Code for my projects. Do I actually need Cowork too, or is it the same thing twice?”

“So Cowork won't replace Claude Code for building my app.”

“I just opened my editor and the name is different and it looks rearranged. Did I break something? Do I have to do anything?”

“So the agent I was using is still there?”

“ok dumb question but if the free plan is just an editor, what am I actually paying $20 for?”

“so the thing in every demo is the paid one. is $20 a lot for that?”

“The article says the database autosuspends after 5 minutes of inactivity. What's the catch with that for a small app — is sleeping a problem?”

“So routine inactivity sleeping and the 'suspended' error at peak are two separate things?”

“Dumb question but if I export my code and leave the app builder, do I lose the live preview and the hosting that made it feel easy in the first place? Or does Claude Code/Cursor give me that too?”

“Okay so once I open the folder in Claude Code, what's the very first thing I even type? I don't want to break it.”

“Dumb question — it keeps saying "Builder ID, not an AWS account." Does signing up mean I'm on the hook for AWS cloud bills somehow?”

“Okay good. So what do I actually get for free without all that?”

“Dumb question — the article keeps comparing the built-in backend to Lovable using Supabase. As a beginner I don't even know what that means. Why would I care that Base44 doesn't use Supabase?”

“Got it. So if I just want users to log in and save stuff without learning a second tool, that's the case for Base44?”

“Dumb question but the review says everything lives inside Blink's platform and auth is native. If I build something on Blink, do I actually get my code, or am I just renting an app I can't move?”

“Got it. So for a quick demo it's fine, but for something I'd keep long-term I should check that first?”

“Dumb question but if my whole app lives in a browser tab, what happens if I accidentally close it before I deploy? Does everything I built just vanish?”

“Okay that helps. But what if I want a copy on my own laptop just in case, before I trust the account to hold it?”

“Dumb question — when it says I hit the wall and get "slower, less capable responses" until the window rolls off, does that mean my actual project or files get messed up? Or is it purely the responses that degrade?”

“Okay good. So how do I even tell I'm getting close before that happens?”

“If I'm brand new, what should I actually install first to learn what a terminal agent does without paying?”

“And switching later means redoing setup?”

“Total beginner here — the setup section says Aider needs 'environment variables.' If I have no idea what that means, am I just locked out of Aider?”

“And Cline really skips all of that?”

“Total beginner here and I'm already confused — the article says the VS Code extension is literally titled 'Windsurf Plugin (formerly Codeium)' and the Neovim repo got renamed too. Did Codeium get discontinued or rebranded into something else? Am I even installing the right thing?”

“Got it. So if I'm not trying to switch editors, I want the plugin and not the whole Windsurf editor?”

“Dumb question, but if autocomplete is firing inline suggestions the whole time I type, doesn't every keystroke burn API tokens? That sounds like it could get expensive fast with BYOK.”

“Okay that's reassuring. So the free Gemini tier is the safest way to dip in without a surprise invoice?”

“Dumb question maybe, but what if Kiro writes a spec and I don't agree with part of it? Am I stuck with whatever it decided, or can a beginner actually push back on the plan?”

“Okay that's reassuring. And if I change my mind later in the project?”

“Dumb question, the article keeps saying React frontend and FastAPI backend. Do I actually need to know what those are to use this, or can a total non-coder just describe the app?”

“So if I'm a beginner still figuring out what I even want to build, is this the tool I should start on?”

“Dumb question but what if I already messed something up BEFORE I ever made a commit? The article is all about committing before the risky prompt, but I only found this after my refactor already nuked a file. Is it just gone?”

“Got it. So step one is literally just make that first commit and stop worrying about the past one. Painful but fair.”

“Total beginner question — if I'm using the free tier and Copilot is reading the files open in my editor to make suggestions, is it doing something with my code? I don't really know what I'm agreeing to.”

“Okay, so the open-files thing is just how it gets context, not a settings I picked?”

“Dumb question but the article keeps saying 'BYOK' and 'API key' — for Copilot I just sign in with GitHub and it works, so why does Cline make me do this extra key thing at all?”

“So which one do I literally install first if I just want to see what AI coding feels like today?”

“Total beginner question: the article says it can "run shell commands" autonomously. That sounds kind of scary on my own machine. Does it just do whatever it wants, or do I get to see what it's about to run first?”

“Good. And if the terminal still feels like too much, is there a way to use it without typing commands at all?”

“It says Lemon Squeezy got acquired by Stripe in 2024. Does that mean if I build on Lemon Squeezy now it's going to disappear and I'll have to redo everything?”

“Okay so it's safe to start there today, just keep an eye on it.”

“Dumb question but the roadmap says pick ONE tool and go. What happens if I get to Stage 2 and realize I picked wrong? Do I have to start the whole app over in a different tool?”

“Okay that actually makes the Git stuff feel less like busywork. So getting to GitHub early is kind of the escape hatch.”

“Honestly I don't even know if it's the AI's fault or mine. How can a total beginner tell whether the code broke or I just clicked something wrong?”

“But what if I made like five changes in the same chat and now I have no idea which one did it?”

“Dumb question: if the .env file is gitignored and never goes to GitHub, how does my teammate get the keys when they clone the repo? Does the file just... not exist for them?”

“So I'd basically commit a fake version with empty values so people know which names to fill in?”

“Dumb question, but when the article says paste the FULL error message, not a summary, doesn't that mean copying my actual code and traceback into ChatGPT? Is that a problem if it's work code?”

“Got it. So the error text matters more than the code around it?”

“Dumb question but if Lovable already builds my whole app, why do I need Supabase at all? Isn't the builder the backend?”

“So Supabase is the part that actually remembers stuff after I close the browser?”

“Dumb question but where do I actually find the file and function names the article keeps saying to include? I'm so new I don't even know what my files are called yet.”

“So I could just paste the error and let it find the file? The article's error example does that, right?”

“Dumb question, but where does this file even go? Repo root? And if I have a CLAUDE.md in a subfolder too, do they fight each other or stack?”

“Got it. So root file = always loaded, path-scoped stuff = loads only when relevant. That clicks now.”

“Dumb question but if I hit the 6,000 completions or 240 chats in a day, what actually happens? Does it just stop, or do I suddenly get charged?”

“Oh. So it's not really a 'just start with it for free' situation anymore?”

“Dumb question, but if Jules clones my repo into Google's cloud to work on it, does that mean my private code is sitting on Google's servers? That makes me a little nervous about using it on a private repo.”

“Got it. So for anything sensitive I'd lean local, and Jules for the stuff I don't mind running in the cloud.”

“Total beginner question: what if the requirements spec Kiro generates is wrong, or I change my mind halfway through? Am I stuck with whatever it wrote at the start?”

“Okay but what about changing my mind after the code already exists? That's the part that scares me.”

“Dumb question but if both tools use the same GGUF models, do I have to download the model twice if I install both? Like does my llama3 download work for both?”

“Got it, so two copies. That's a lot of gigs for the same brain.”

“Total newbie question — if I'm on the free tier and run out of credits halfway through building my app, does my app just vanish, or is it saved somewhere?”

“Oh good. And is there a way to make a backup so I don't lose my work if something goes wrong?”

“Dumb question, but after I pull a few models with ollama run, where do they actually live and how do I get rid of one I don't want anymore? I'm worried about filling up my drive with 5 GB downloads I forgot about.”

“Got it, that helps. So nothing leaves my disk automatically, I just have to clean up myself?”

“Naive question: when it reads my project files to do a task, is all my code getting sent to OpenAI?”

“Is there a mode where it doesn't just go off and do stuff on its own?”

“Total beginner here. The article says it's not an IDE and doesn't replace Cursor, so what would I actually use it FOR if I'm not coding on my phone?”

“Got it, so it's for triggering tasks remotely, not writing code with me. Is Telegram the easiest way to start?”

“Dumb question but if I split one prompt into three, isn't that just slower? Feels like I'm doing three round trips instead of one.”

“Okay, so the time I'd lose chaining I'd lose anyway debugging. Got it.”

“Dumb question, but the article says Void sends my request 'directly to the model provider' with my own key. If I use a Claude or Gemini key, isn't my code still going to a company's server? How is that more private than Cursor?”

“Got it. So local models = nothing leaves, cloud key = only the model provider sees it. That makes sense.”

“Dumb question, but if I just keep starting fresh chats every task like fix 1 says, am I going to lose track of what I already decided? Feels like I'd be retyping the same setup forever.”

“Got it. So the summary is just 'what am I doing right now' and the file is 'everything that's always true.' That clicks.”

“Total beginner here. When the article says "start a fresh chat," does that mean the AI forgets my actual code too? Like does my project disappear, or just the conversation?”

“Oh okay. So the summary is basically me reminding it of stuff it can't re-read on its own, like why we chose something?”

“Dumb question, but if it lives on a server, how does it actually end up in my WhatsApp? Do I install an app or something?”

“Got it. And I don't have to set up all 20-something apps, right? Just the one?”

“Dumb question — if each agent works in its own sandboxed copy (the worktree thing), how do the changes from two agents end up back in my actual project? Do they just merge automatically?”

“Got it, so as a beginner I don't really need to think about it yet?”

“Dumb question — everyone keeps saying "just bring your own API key" like it's free. If I have to pay the API per token, how is that the cheap option versus the $10 subscription?”

“So as a beginner doing small projects, BYOK is probably the lower bill?”

“Dumb question, but if I install one of these and then change my mind, is there an undo? Or am I stuck editing that JSON file by hand?”

“Okay that helps. But the GitHub one wants a PAT token and that part terrifies me. Is that going to be a whole ordeal?”

“I've never touched an API key in my life — is the BYOK step going to wall me out before I even start?”

“Everyone online makes the terminal sound cooler and more pro. Am I cheating or falling behind if I stick with the clicky editor instead of learning the command line one?”

“So is there ever a real reason to switch, or is the terminal just hype dressed up as progress?”

“I have literally never touched Git in my life. The whole 'initialize a repo and push to GitHub' step sounds like the part where I get stuck for two hours. Is there a way to skip GitHub entirely for a first deploy?”

“And if I do try the GitHub route and it yells at me about credentials, what then? That error usually ends my evening.”

“Dumb question — this is all for Claude Code in the terminal, right? Does the same setup work if I'm just using Claude in the normal app/browser?”

“Okay. And the npm install line at the top — that's installing Claude Code itself, not the Figma part?”

“Dumb question but if everything lives in Replit's cloud, what actually happens to my app when I close the browser tab? Does it just stop existing?”

“Got it, so the difference is between it being saved versus it being reachable. That clicks now.”

“Dumb question, but if the free tier only gives 10 fast requests a month, does that mean after 10 messages the whole editor stops working? I'm scared I'll get locked out mid-tutorial.”

“Oh okay, slow requests sound like the move then. Is there a catch to those?”

“Dumb question maybe, but if I have a global file AND a project file AND one in a subfolder, and they say different things, which one wins? Does the most specific one override the others?”

“Oh okay, so it's not winner-take-all, it just adds them up. That's less confusing than I thought.”

“Dumb question — if Context7 fetches live docs every time, doesn't that make every prompt way slower? Or am I overthinking it?”

“Okay so I don't have to slap 'use context7' on literally every prompt then?”

“Honest question, is it cheating if I never read the code the AI writes? It feels like I didn't really build the thing.”

“Okay but then how do I tell my friends I made it without feeling like a fraud?”

“I'm brand new and don't get the difference. When am I supposed to use Cascade versus just typing and letting it autocomplete?”

“And that agent part is what runs out, right? How do I avoid burning it by accident?”

“i just have a folder of html and css files that the AI builder spit out. whats the absolute least amount of steps to get a link i can text my friend?”

“wait no signup at all?? whats the catch, does the link disappear?”

“total beginner here. if I just want SOMETHING live today with zero friction, which do I open and which do I avoid?”

“and the database thing everyone mentions — do I have to care about that on day one?”

“Dumb question but if the skill already activated fine the first couple times, why would it suddenly stop partway through a session? Did I break something?”

“Okay that's a relief. So the hook is basically a sticky note that says check your skills first, every time?”

“Dumb question, but if I set up a format hook and the formatter command itself errors out, does the whole thing just break? I'd rather know before I touch my settings file.”

“Okay, so when does exiting with an error actually stop Claude from doing something?”

“Dumb question, but the article keeps saying "tell your AI tool" and pasting these little prompt blocks. Do I type those exactly as written, or are they just examples I'm supposed to reword?”

“Got it. And the {name} part with the curly braces — that's a placeholder, right, not something I literally leave in the URL?”

“Honest question from someone totally new: if the installer keeps updating itself in the background, do I ever have to reinstall or run anything again to stay current?”

“Oh nice. So after the very first install, the only command I really need is just typing claude to start it?”

“Dumb question but is the terminal thing the same as that black hacker window in movies? I'm a little scared I'll mess up my whole laptop by typing the wrong thing.”

“Okay that helps. So if I see a scary red error, what does that actually mean for me?”

“I can barely write CSS yet. If this thing hands me JetBrains Mono and green CTAs, am I just stuck because I cant turn that into actual code?”

“Okay but what stops me from making the classic beginner mistakes everyone warns about?”

“If it already acts like an agent by default, why would a beginner ever need to mess with the /agents command at all?”

“So the practical change is really just how I phrase things to it?”