1DesignToolDownload

Automation · since v1.6.0

Design MCP server

MCP is how your agents ask the studio to design: start a project, send a brief, read the gate verdict, pull an export — everything goes through grants you gave, logged in the activity view.

DownloadRead the docs

Why it matters

Design automation usually means screenshots and API scraping. With the studio as an MCP server, an agent can treat design as a capability: 'make this repo a landing page' becomes a call, not a copy-paste — and the grant list keeps the authority honest: agents do only what you allowed, and the activity log shows exactly what they did.

How it works

  1. 1Enable the serverSettings → MCP exposes the studio's server; an MCP-capable client (Claude Code, Cursor, others) adds it like any tool server.
  2. 2Grants define the reachAccess is scoped by grants — server_info reads, project creation, turn dispatch, export pulls. An agent gets only the tools you granted.
  3. 3Agents call design like a functionBrief a project, wait on the gate, fetch the artifact — the same loop a human drives, callable.
  4. 4Watch what they didThe activity log records every MCP call with its arguments and result — design automation with an audit trail.

Capabilities

  • The studio as an MCP server: projects, briefs, gate verdicts, exports as tools
  • Scoped grants — an agent sees only the capabilities you allowed
  • Activity log of every MCP call, arguments and results
  • Works with any MCP-capable client — the agents already driving your projects qualify
  • Same kit, same gate: automated turns get the same verification human ones do

FAQ

What can an agent do over MCP?

Whatever its grants allow — from read-only server_info through creating projects, dispatching turns and pulling exports. The grant list is the boundary.

Is my design work exposed to the agent's cloud?

The MCP server answers the client's calls; what's in the responses — briefs, verdicts, files the agent already produced — is the same material that agent wrote. Nothing extra is sent anywhere.

Can I see what an automated run did?

Yes — the activity log records each call with its arguments and outcome, and the project's normal history shows the turns it produced.

In the docs

Related features

AgentsYour coding agent is the designerClaude Code, Codex, Cursor Agent, Gemini, Devin and eleven more — whichever CLI you trust, 1DesignTool drives it as a designer and holds its output to the same visual bar.Read more →SteeringA design kit the agent has to passThe gate is the difference between 'generated' and 'shipped': kit checks run on every turn, and in Reviewed or Council mode independent critics judge the frozen evidence before the result counts.Read more →AgentsClaude Code design toolPoint 1DesignTool at your Claude Code install and it becomes your designer — writing HTML, CSS and JS into a project folder you own, with a live preview and a design check before anything counts.Read more →

Design with the agent you already have.

Free to try, buy once — the studio and your own coding CLIs do the rest.

Download