Connect an AI assistant to Layer

Last updated: October 11, 2026

Add Layer to Claude, Cursor, VS Code or another assistant that supports MCP, and ask it to generate images, video, 3D and audio in your Layer workspace, with your styles and reference sets, from the conversation you are already in.

Before you start

  • A Layer account. You sign in with it the first time the assistant uses Layer, and the assistant then works as you, in the workspaces you belong to. Generations use your workspace's Creative Units, as they do in the browser.
  • The Layer MCP server address, which every client below uses: https://mcp.app.layer.ai/mcp

Connect your assistant

Claude

  1. In Claude, open Settings, Connectors and click Add custom connector.

  2. Paste the Layer MCP server address and click Add.

  3. Sign in with your Layer account when Claude asks.

Claude Code

  1. Add the server from your terminal: claude mcp add --transport http Layer https://mcp.app.layer.ai/mcp

  2. Run /mcp, select Layer and choose Authenticate.

Cursor

  1. Open Settings, Features, MCP and click Add New MCP Server.

  2. Choose Streamable HTTP, name it Layer, and paste the server address.

  3. Sign in with your Layer account when Cursor asks.

VS Code

  1. Add Layer to .vscode/mcp.json as a server named Layer of type http, with the server address as its url.

  2. Sign in with your Layer account the first time it is used.

Codex

  1. Register the server: codex mcp add layer -- npx -y mcp-remote@latest https://mcp.app.layer.ai/mcp

  2. Sign in with your Layer account the first time it is used.

Windsurf

  1. Add Layer to ~/.codeium/windsurf/mcp_config.json under mcpServers, with the server address as its serverUrl.

  2. Check it works by asking: "What can I do with Layer?"

Grok

  1. Open grok.com/connectors, click New connector and choose Custom (on a team plan, + Add Connector, Other).

  2. Paste the server address and sign in with your Layer account when Grok asks.

Any other client

A client that can launch a command connects through the mcp-remote bridge: configure a server that runs npx -y mcp-remote https://mcp.app.layer.ai/mcp. Setup for Zed, Warp, Gemini CLI, Amp, OpenCode and others is in the MCP setup docs.

Result

Ask the assistant for an asset, for example "Generate four icon concepts for a frost sword in our house style", and it picks a model, estimates the cost, runs the generation in your workspace and shows you the result. Everything it makes is in your Layer workspace, where the rest of the team can find it.

Production tips

  • Name the workspace and project. When you belong to several workspaces, say which one, and which project the work belongs to, so the files land in the right place.
  • Ask for the price first. The assistant can estimate a run in Creative Units before it spends them. Ask for the estimate on large batches.
  • Run your team's workflows. Saved workflows are available to the assistant too. Ask it to list them and run one with your inputs.
  • Build on it in code. For scripts and pipelines, see the API and SDK at layer.ai/docs.

If something goes wrong

  • The assistant says it is not authorized, or asks you to connect your Layer account. The sign-in did not finish or has expired. Reconnect Layer in the assistant's connector or MCP settings and sign in again.
  • Your workspace is missing. You signed in with a different Layer account than the one invited to the workspace. Disconnect, then sign in with the right account.
  • A generation fails for lack of Creative Units. The workspace has run out. See Manage your subscription and buy Creative Units.

Related