Skip to content
Mistflow

Quickstart

Get from zero to a deployed app in three steps. Total time: about 2 minutes.

Prerequisites

Section titled “Prerequisites”

  1. Install Mistflow

    Terminal window
    npx -y mistflow-ai install

    The installer auto-detects your editors and configures them.

  2. Describe your app

    Open your AI editor and describe what you want to build:

    With mistflow, build me a SaaS landing page for a developer
    analytics tool called "DevPulse". It should have a hero section,
    feature grid, pricing table with 3 tiers, and a waitlist signup form.

    Mistflow takes it from there, it plans the app, scaffolds the code, builds each feature, and deploys it. You just watch and approve.

  3. See it live

    Within seconds, you’ll get a URL like:

    https://devpulse.mistflow.app

    Open it. Your app is live. You can now:

What just happened?

Section titled “What just happened?”

Behind the scenes, your AI editor worked with Mistflow through four stages:

graph LR
    A[You describe the app] --> B[Plan]
    B --> C[Scaffold]
    C --> D[Build features]
    D --> E[Deploy]
    E --> F[Live app]

Every step is recorded. Open the Mistflow dashboard to see the full build history, deployment logs, and provenance trail for every feature.

The 14 MCP tools

Section titled “The 14 MCP tools”

Mistflow ships as a single MCP server (@mistflow-ai/mcp) with 14 discrete tools. Your AI editor calls them directly — there’s no separate CLI binary to install or keep in sync. Long-running work (mist_install, mist_build, mist_qa, mist_deploy, mist_plan) uses fire-and-poll: the first call returns a jobId with status: "running"; subsequent calls with that jobId return current state in under a second. The 60s MCP timeout never bites.

See How It Works for the full tool list and Your First App for the end-to-end chain.

Next steps

Section titled “Next steps”