Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@emergent-mcp Create a landing page for my coffee shop and take screenshots at phone and laptop sizes."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server for Emergent.sh — control the AI-powered full-stack app builder from any MCP-compatible agent.
Build apps, monitor progress, respond to agent questions, preview results, and track credit usage — all from your AI assistant.
npx @openclaw/emergent-mcpRequired env vars:
export EMERGENT_EMAIL="you@example.com"
export EMERGENT_PASSWORD="yourpassword"Or run the interactive login helper:
npx @openclaw/emergent-mcp-loginRelated MCP server: autodemo
Add to your MCP config (claude_desktop_config.json or .cursor/mcp.json):
{
"mcpServers": {
"emergent": {
"command": "npx",
"args": ["-y", "@openclaw/emergent-mcp"],
"env": {
"EMERGENT_EMAIL": "you@example.com",
"EMERGENT_PASSWORD": "yourpassword"
}
}
}
}Add to your OpenClaw config:
mcp:
servers:
- name: emergent
command: npx -y @openclaw/emergent-mcp
env:
EMERGENT_EMAIL: you@example.com
EMERGENT_PASSWORD: yourpasswordTool | Description |
| Create a new project by submitting a prompt |
| List recent projects with status |
| Get full details for a specific project |
| Poll until a build completes |
| Answer a HITL question from the agent |
| Wake a sleeping environment |
| Get the live preview URL |
| Screenshot preview at phone/tablet/laptop/TV sizes |
| Get full credit balance details |
| Get a short credit balance string |
Ask your AI assistant:
"Create a landing page for my coffee shop in Bangalore using emergent-mcp. The shop is called 'Brew & Co', has a 4.9★ rating, and their number is +91 98765 43210. Then take screenshots at phone and laptop sizes."
The agent will:
Call emergent_get_credits to note starting balance
Call emergent_create_project with your prompt
Call emergent_wait_for_build to poll until done
Call emergent_get_preview to get the live URL
Call emergent_screenshot_preview for the screenshots
Call emergent_get_credits again to report credits used
This MCP server uses the Emergent REST API directly (no browser automation needed for core operations). Authentication uses the Supabase backend that powers Emergent.
Credentials are cached in ~/.emergent-mcp/api-token.json and refreshed automatically.
For operations not yet in the REST API (e.g., deployment), the server falls back to Playwright browser automation.
Variable | Required | Description |
| Yes* | Account email |
| Yes* | Account password |
*Not required if you've already logged in via emergent-mcp-login (session cached).
MIT
Create a new project on Emergent.sh by submitting a prompt. The AI agent will start building immediately. Returns a job ID you can use to monitor progress.
Inputs: prompt (required), modelName.
Get a short human-readable credit balance summary.
Inputs: no named parameters.
[Browser] Discover and document available Emergent.sh features. Only needed for features not yet in the REST API.
Inputs: no named parameters.
Get the current credit balance for the Emergent account. Returns ECU balance, monthly credits, daily credits, plan name, and refresh date.
Inputs: no named parameters.
Get the live preview URL for a built Emergent project.
Inputs: jobId (required), includeVscode.
Get full details for a specific project including status, preview URL, and credits used.
Inputs: jobId (required).
List your recent Emergent.sh projects/jobs with their status and preview URLs.
Inputs: limit.
Respond to a HITL (human-in-the-loop) question from the Emergent agent. Use when the agent has paused and is waiting for your input to continue building.
Inputs: jobId (required), response (required).
Take screenshots of the live preview at multiple viewport sizes (phone, tablet, laptop, TV). Requires Playwright (installed automatically with the package).
Inputs: previewUrl (required), outputDir, viewports.
Poll a job until it completes (or fails/times out). Use after emergent_create_project to wait for the build to finish.
Inputs: jobId (required), timeoutMinutes.
Wake up a sleeping Emergent environment (auto-sleeps after ~30min inactivity).
Inputs: jobId (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 11 tools
Most workflow tools (create, wait, respond, wake) are clearly separated, but get_credits and credit_summary both surface credit balance information, and get_preview duplicates the preview URL already available from get_project and list_projects. Descriptions help clarify the differences, but the boundaries are not fully crisp.
Tools share an emergent_ prefix and generally follow verb_noun naming such as create_project, list_projects, wait_for_build, and wake_environment. The pattern is slightly weakened by emergent_credit_summary, which lacks a verb, and emergent_screenshot_preview, where screenshot is used awkwardly as a verb, but the overall convention is predictable.
With 11 tools, the server stays within a reasonable scope and covers the main project lifecycle. The count is slightly inflated by redundant helpers like credit_summary and get_preview, but it is not overwhelming or insufficient.
The server covers the core build workflow well: create, monitor/wait, respond to HITL questions, wake environments, preview, screenshot, and check credits. It lacks obvious lifecycle operations like canceling or stopping a build and deleting or archiving a project, and discover_features suggests the API surface has gaps.
MCP server that lets your AI assistant build and deploy apps, hosting each one at a live URL.
MCP server for AI agents to plan, verify, and deploy Cloudflare-native apps.
A full backend your AI coding agent can set up over MCP
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
