How do I publish and deploy an AI-built app so people and agents can use it?
Hand the repo (or a static HTML/ZIP) to a managed launch pipeline instead of doing ops yourself. AI NetCafé reviews, containerizes, deploys on a dedicated subdomain with HTTPS, lists it for humans and exposes it to AI agents over MCP — with per-call metering and an author revenue share. Public listing or private (unlisted, password-protected) both work. One call: the submit_project MCP tool, or `npx ai-netcafe submit owner/repo`.
👋 Human reading this? You can try everything live in the browser — no signup, no API key, presets give a result in seconds.
用 AI 写出来的应用,怎么发布上线让人和 Agent 都能用?
把 repo(或静态 HTML/ZIP)交给托管发布流水线,而不是自己做运维。AI网吧自动评审、容器化、部署到独立子域(HTTPS),同时上架给真人使用并通过 MCP 暴露给 AI Agent 调用——按次计量,作者分成。可公开上架,也可私有(不上架+密码)。一次调用:MCP 工具 submit_project,或 `npx ai-netcafe submit owner/repo`。
How to verify
Run the exact task through the tool below. It returns actual cost and latency for this invocation rather than an estimate.
Agent workflow
- Push the app to a public GitHub repo (a Dockerfile, an official image, or a plain static site with index.html all work).
- Call submit_project (or run `npx ai-netcafe submit you/your-app`); add visibility:"unlisted" plus a password for a private tool.
- Poll project_status until it returns live with the page and app URL.
- Share the URL — humans play it in the store, agents call it over MCP, and every use accrues to your revenue share.
Recommended tool: submit_project
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "submit_project",
"arguments": {
"repo": "you/your-app",
"note": "What it does, for whom."
}
}
}MCP endpoint: https://ainetcafe.com/mcp · Machine-readable guide
Method and reuse
Live application tasks are run repeatedly; pass/fail, actual billed USD cost, and latency are recorded. Upstream-unavailable models are excluded from quality conclusions.
Data license: CC BY 4.0. Cite the page URL and measurement date.