Use cases · AI apps and agents

Built in the editorRunning in production

Ship the AI app you already built. Inspect deploys, read logs, and keep writing.

Connect your editor Open AI workflow

prompt

inspect the storefront deploy and review the logs

mcp

deploylogsvariablesdatabases

live

storefront.up.dflow.sh

Before

The editor stops at code.

Generation is fast. Production still has to be invented from consoles and scripts.

app/page.tsx

export default function Page() {
  return <Storefront />
}

no production path

Infra, secrets, and the URL are still a side project.

MCP

The editor ships. Not the cloud account.

The agent can deploy, read logs, and attach data inside the workspace role.

tools/call

create_deployment

paramsstorefront-web · prod

resultqueued → success

Trigger a dFlow release from the editor.

Access

OAuth into the workspace.

Cursor, Claude Code, Copilot, Codex, and Antigravity sign in. Any MCP client can use the same server.

Cursor

.cursor/mcp.json

Claude Code

.mcp.json

GitHub Copilot

.vscode/mcp.json

Codex

Codex MCP

Antigravity

Antigravity MCP

mcp

Any MCP client

streamable HTTP

Control

Role scope instead of a provider token.

The agent sees deploy, logs, and variables. Not the whole cloud account.

provider token

cloud:*

A raw provider key is more than a routine deploy needs.

workspace role

deploy · logs · vars

OAuth plus workspace role. Scope travels with the user.

agentic

Build with AI.Ship with production context.

mcp.dflow.sh

Use dFlow when AI-generated software needs a real deployment path, not another cloud puzzle.

Connect your editor