“I'm glad they changed things up a bit because I was honestly getting close to being done with Replit. After a few months of extra charges, it was starting to feel like I was getting price gouged just for normal usage.”
“Currently in the middle of a big project and I recently noticed that "Free Mode" has been added. I was curious to see everyone's take on free mode and hear if it's at all worth running. My main concern is if it'll jank up some code if I give it tasks.”
“Preserve the current production app, pick one recurring failure that support has already failed to resolve, and reproduce it on a copy with a clear pass/fail test. Then compare three things: whether Replit can fix it cleanly, whether an outside tool can patch it without collateral changes, and what each path actually costs.”
“I only use Replit ONCE on FREE to create a Base Template for my App.. then connect to ExpoGo, also ONLY ONCE, and connected my Github with Both. NOW I ONLY WORK USING GITHUB ACTIONS..”
“you can ask Replit Free mode to plan and lay down the infra.. Then ask it to build as much as it can. Then you can give the same plan to claude in the shell ask it - 1.”
“I've got a self-hosted hardware config based on Apple Silicon, but haven't found an environment that can replace Replit in terms of connections and straightforward harness.”
“Once that is done easily via replit then it's not hard for a technical guy to go in and improve the app even within replit. We've been doing it for a while and it's working well!”
“You’ll save a lot of money because OpenAI is significantly cheaper than Replit, especially if you’re building a medium-sized app. For a small app that can be completed in a few days, Replit is sufficient, as you’ll likely exhaust their monthly credit anyway.”
“check ur invoices, replit tends to double charge customers because of their rolling bills. I complained big time and they gave me a refund (kind of - as credits).”
“I was in a similar spot, was spending over $1k/month with Replit and now I'm spending around $200/month for a better set up with similar function to coding but way more integration capabilities.”
“Significantly stupider and bolder. Goes off on its own and makes things I never asked for, no update or chat from it when in the past it always explained what it was doing. This is garbage now honestly.”
“Replit is truly the worst option for 85% of the people using it and no matter how many times I used to read how bad it was I just stayed, mainly because it was easier than migrating.”
“I’ve noticed that every time I ask Replit to add something common—like authentication, billing, or an admin page—it builds it from scratch again. Sometimes it works. Sometimes it takes several prompts to fix what changed.”
“If you don't pay the $100 a month tier then you will just have access to Replit controlled Clerk which means you have no access and is not great on a production app since the login splash screen says logging onto Replit instead your app/webpage.”
“I've been using Claude inside Replit for awhile and probably once a day, the shell closes and my session crashes, losing all the progress that wasn't committed.”
“For someone who’s built a production app on Replit but no experience with any other type of coding, is Claude Code suitable - or more for experienced coders? Thanks.”
“I think MCP is more useful it does all of this plus more. [https://docs.replit.com/platforms/mcp-server](https://docs.replit.com/platforms/mcp-server)”
“I'm assuming no considering I've uploaded a .zip file website to Gemini, Replit, and Claude before, and Gemini was the only one who got the recreation totally wrong.”
“Claude Code will modify the code in your GH Desktop repo... then you will push/upload it to GitHub Online... and from there pull it down into Replit using the "shell" utility, then republish the Replit app, which updates the app with the new code. Once you get the hang of it ...”
“Here's a full walkthrough I made on how you can push the code up to GitHub and then pull it down locally and work on the app locally and get around any of the Replit bugs (because you can set up your own environment), and then it even shows you how to publish it.”
“I don’t think it is only credits. Credits hurt because they make failed edits visible in dollars, but the deeper issue is trust. If the agent burns money, breaks something, and you cannot easily inspect, export, or control the environment, people start feeling trapped.”
“What I do is install the GitHub plug-in in ChatGPT. Start a new project and give that project specific instructions to only access the repository for your app to always ask for permission before making any changes, and never make edits directly to the main branch so always create a new PR.”
“There should be a Replit integration inside ChatGPT where you can just let ChatGPT work inside Replit (but you give it instructions inside the ChatGPT UI).”
“the production data problem is likely the thing you should isolate. Look at your production environment variables, secrets, database permissions and api connection one by one before you start debugging the agent. This usually helps you figure out if the issue is with replit with the deployment or, with the data connection.”
“I have been using Replit for about 6 months, and although I have been able to build working applications, I must warn everyone that it is full of bugs. I have had to reach out to Replit support multiple times for publishing issues, agent malfunctions, and, most recently, an agent that cannot reach production data.”
“my team uses replit enterprise. however, team is not able to make commits to github enterprise from replit because the GitHub connector is broken. And Replit support can't fix that. Just having SSO can't justify an enterprise level subscription”
“I've used it for about 6 months and been extremely successful most of the times. It has really improved My clarity of thought because I have to know order of events and dependencies prior to building. In fact non-UX changes are usually slow multiple iterations of prompts.”
“For anyone who has been burning through credits using full-power agent models for routine tasks: moving day-to-day context generation and coding tasks to Claude Code alongside Replit's free tier is an absolute game-changer. I should have integrated Claude earlier (like a month ago!).”