AI AGENT DEPLOYMENT GUIDE

MCP deployment hosting for AI coding agents

Yeeted is a hosting platform for AI-built apps. Your coding agent operates it through MCP. Instead of translating an agent’s work into a separate CI or cloud-console workflow, connect the agent to Yeeted, deploy a preview, verify it, and promote that exact build when it is ready.

Connect an MCP client

{
  "mcpServers": {
    "yeeted": {"type":"http","url":"https://mcp.yeeted.ai/mcp"}
  }
}

On first use, the client opens a browser to authorize the connection. During the Yeeted beta, the underlying account requires an invite code and a verified email; MCP authorization does not create an account.

Use a preview-first deployment loop

Ask the agent to inspect its available tools and project context.
Deploy the app to a development preview.
Run a smoke test and inspect logs or diagnostics if needed.
Promote the tested version to production (Small Business plan and up, after the beta), or restore app code from a named checkpoint or retained git hash.

Yeeted auto-detects common web stacks and can manage databases, logs, secrets, and domains. Applications should listen on 0.0.0.0 and honor $PORT.

A git-hash rollback restores app code only; database changes are not automatically rolled back. A named production checkpoint with a captured database snapshot can also restore supported production data except Redis on plans with snapshots if the restore succeeds. Check the rollback result. Use an independent backup for code and data. Yeeted is in beta and is intended for testing and development rather than mission-critical workloads.

Next steps

Read the Yeeted AI deployment documentation, or use the focused Claude Code and Cursor guides.