CleanSlop
Ship vibe-coded apps with confidence
31 followers
Ship vibe-coded apps with confidence
31 followers
CleanSlop helps you ship vibe-coded TypeScript apps with confidence. Sign in with GitHub, connect your repos and run a scan. It maps the codebase (call paths, module dependencies, background jobs, data stores, external services) and surfaces the problems that matter: things that work in a demo but break with real users, like unbounded work, slow operations blocking a request, and jobs that quietly get dropped. You get a short list of real risks worth fixing, not hundreds of findings to triage.
This is the 2nd launch from CleanSlop. View more
Agentic CleanSlop
Launching today
Paste one line into Claude Code, Codex, Cursor, Grok Bot or Muse. Your agent authenticates you with Stripe Link (no sign-up, no password), scans your vibe-coded TypeScript/Node.js app for what breaks with real users, then fixes it. More scans? Approve in Link.
Free Options
Launch Team / Built With
ArticosTest your messaging on your audience, right inside Claude or ChatGPT
Promoted
Hey Product Hunt 👋 Swaroop here, one of the two people building CleanSlop.
A month ago we launched CleanSlop here: a second look at the app you built with AI, telling you in plain English what could break once real people use it, and what to fix first.
One thing kept bugging us: you'd build in Claude Code, Codex or Cursor, then have to leave your agent for our website to sign up, run a scan and read the report. The blocker was authentication. An agent can sign up for things, but a service can't tell whether a real, accountable person is behind it, or who that person is.
That changed with Link Agent Identity, Stripe's new identity product for agents, and we're one of the first companies to integrate it! Your agent presents Link's attestation that a real person stands behind it, along with their Link-verified email, right from the CLI. No form, no password, no trip to a website.
So today, CleanSlop is fully agentic. You paste one line into your agent:
and it does the rest:
Authenticates you with Link, through that attestation. Nothing to approve.. No sign-up form, no password.
Scans the folder you're in. Every finding is reviewed against your code before you see it, so you get a short list, not 200 warnings.
Fixes what's worth fixing, then rescans to check the fixes held.
Payments go through Link too. When you run out of scans, your agent offers a credit pack or a monthly plan, and you approve it in Link. Your agent never sees your card, and that approval is the only time you touch anything.
It works in coding agents like Claude Code, Codex and Cursor, and in the new personal agents such as Grok Bot and Muse.
What's supported today:
TypeScript/Node.js apps, sharpest on Next.js + Supabase.
Link identity in the US and Canada for now.
Anywhere else, your agent can sign in with GitHub instead.
🎁 Exclusively for Product Hunt: tell your agent:
We'd love to hear where it gets confusing or where your agent gets stuck. We're here all day 🙏
Hey Product Hunt!
Anomit here, co-maker of CleanSlop along with Swaroop.
It has been an insanely productive 4 weeks since our last launch here as we incorporated our early users' feedback on almost a daily basis. Which made it necessary for us to let the community in on a leaner, faster and completely agentic CleanSlop with this release. I will attempt to summarize them below, but the best way to keep up with our rapid iterations is to become a regular user and have the weekly changelogs delivered straight to your inbox.
A completely agentic workflow backed by Stripe's Link Agent Identity, which attests for the genuineness of agents that use CleanSlop's services.
The only human interaction involved here is if you ever need to approve a spending request when your pack credits run out.
For non-US and Canada users, we still support sign-ins with GitHub that can be used to upload code from your local machine, without the need for a GitHub repo.
We have significantly re-trained the scanning engine to surface issues truly worth being concerned about, while reducing scan times by upto 99% (you read that right)!
cleanslop.com itself is a self-contained agentic resource now. All you need to do is point your Claude, Codex, Muse, Grok Bot or any other agent or personal assistant at it with a single line of a prompt from within your code tree. The resources are structured in such a way that your agent or personal assistant will pick up the right pathway.
We have a cleanslop CLI as well as a MCP server. For human readable guides on their usage, visit our Agents or MCP resources. Your agent picks them either way if you just use the single line prompt.
We are all ears eagerly waiting for our early users' to give this improved CleanSlop a spin and leave us feedback, questions, criticisms -- anything that you would like us to be aware of. We will be monitoring the comments and our inbox throughout.
Happy cleaning the slop out of your codebases!
The name made me laugh, which is already a win on PH. Running the scan from inside the agent instead of a separate tool makes sense. Does it handle auth flows or is that still manual?
@jean_bonnenfant_tinct It handles it for you! Only requires an approval if you don't have Link CLI on your machine. Give it a go and let us know :)