works with any AI coding agent · via MCP

Hosting for AI-built apps.
Run by your coding agent.

You vibe it. We ship it. 🚀

Yeeted is a hosting platform for AI-built apps. Your coding agent deploys and manages your app, databases, logs, domains, and rollbacks from the conversation.

your AI agent
Deploy my recipe app to production
Deploying via Yeeted…
Built your app   Database ready   Tested on a copy of prod
Live at recipes.yeeted.ai 🎉

Three words: “deploy my app”

01 - ask
Your agent calls Yeeted

Claude, Cursor, or any MCP-enabled agent sends your code straight to us. No terminals, no config files.

02 - try
Test on a safe copy

Every deploy lands in a test space first - with a snapshot of your real data, so nothing breaks for real users.

03 - ship
Promote when happy

Say “make it live.” We swap it into production instantly - and one word rolls it back.

built for AI development loops

The deploy target your agent can drive alone

Agents don't read dashboards - they read tool results. Every Yeeted answer is designed to be acted on by a model: structured, honest, and never blaming your code for our problems.

blameless failure classification
Your agent always knows whose bug it is

Every failure carries cause_category and self_fixable — with file:line when it's the code. A capacity miss says "platform, don't touch your app." A stack trace says "fix this line." No more agents rewriting healthy code because a build queue hiccuped.

deploy → test → diagnose → promote
The whole loop, as tools

validate dry-runs before shipping, smoke_test probes the live app, diagnose returns one internally-consistent verdict with a recommended action, and rollback undoes a bad ship by name. An agent can run the entire cycle unattended.

safe to retry, safe to batch
Idempotent by content, paced by headroom

Re-sending identical code is a no-op, not a second build — retry loops can't stampede. project_context reports live capacity headroom, and queue_on_capacity parks a batch deploy until a slot frees instead of failing it.

a memory for your loop
Versions, checkpoints, and diffs

Every deploy is a numbered version with stored source. whats_changed diffs two versions, checkpoint names a known-good state, and new errors are flagged new since last deploy — so the agent knows whether to fix forward or roll back.

Read the agent-loop guide →

Under the hood

For the curious: your history is the source of truth, immutable builds are what actually run.

YOUR SIDE
AI agent
Claude / Cursor / any MCP client
API SURFACE
MCP server
deploy · promote · rollback · logs · db
CONTROL PLANE
Hidden git repo — full history, “roll back to Tuesday”
Buildpacks — auto-detect any stack
Image registry — immutable, instant rollback
Test environmentmicroVM
Its own database, seeded from a snapshot of your production data — test against real-shaped data (Business and up)
✓ health checks↓ promote same image
Productionblue/green
Zero-downtime swaps · 1→N replicas by plan · multi-region + failover on High Availability · one command to roll back
MANAGED FOR YOU — CREATED AUTOMATICALLY WHEN YOUR APP NEEDS THEM
SQL databases — Postgres, MySQL & MariaDB
MongoDB — document store, on demand
Redis — cache & queues
Logs & metrics — readable by your agent, so it can fix bugs itself

Plans that grow with your app

More resources, more resilience, safer testing — as you need it.

One flat monthly price. No hourly billing. No bandwidth charges. No per-seat fees. No surprise overages. Your bill is the number on the card — every month, no matter how much your app is used. Bandwidth is unmetered under our fair-use policy: heavy sustained traffic may be throttled, never billed.

Billed monthly · Cancel any time

Try it for free. 5 deployments, 5 projects, no cost. Start building →
Developer
$10/mo
5 projects · 1 live preview
Sleeps when idle, wakes on request
A small managed database
Start building
most popular
Pro
$20/mo
30 projects · 3 running at a time
Managed databases, two engine types
Room to build — upgrade to Business to go live
Start building
Business
$50/mo
50 projects · 1 production app
Larger 2 GB databases
Everything in Pro
Coming soon
High Availability
$199/mo
50 projects · 2 regions, auto-failover
Everything in Business, more replicas
Cross-region database sync
Coming soon

Every plan includes unmetered bandwidth and builds (fair use applies) — the price above is the whole bill.

Need dedicated capacity, SSO, or a custom SLA? Talk to us about Enterprise →

Common questions

Short answers first; the docs go deeper.

What is Yeeted?

Yeeted is a hosting platform built for apps written with AI coding agents. Your agent deploys straight from the conversation you are already having with it, and Yeeted runs the app with a managed database, logs, secrets and domains - you never touch a server or a config file.

How do I deploy an app with Yeeted?

Tell your agent to deploy - literally “yeet it”. The agent calls Yeeted’s MCP tools, your code is built into a container and comes up on a live URL, usually in under a minute. There is no git push, no YAML to write and no dashboard step in the loop.

Which AI coding agents does Yeeted work with?

Any agent that speaks MCP over HTTP: Claude Code, Cursor, Windsurf and others. Yeeted is agent-agnostic by design - it works inside whatever tool you already build with, rather than asking you to switch.

Does Yeeted provide databases?

Yes. Yeeted attaches managed Postgres, MySQL, MariaDB, MongoDB or Redis to your project on request, injects the connection details automatically, and supports snapshots, exports and password rotation - all driveable by your agent.

Can I roll back a bad deploy?

Every deploy is an immutable, named version, so rolling back is one instruction to your agent. Because Yeeted keeps this history for you, it also works if you don’t use git at all - Yeeted is the version history for your app.

Do I need Git to use Yeeted?

No. Yeeted deploys the code your agent has in hand and records every deploy as a snapshot you can name, diff and restore. If you do use git, that works too - Yeeted just doesn’t require it.