View and search Lucid content with AI tools using the Lucid read-only MCP server

Written by:  Morgan T
Last updated:  

The Lucid read-only Model Context Protocol (MCP) server enables AI tools like ChatGPT, Claude, and Microsoft Copilot to securely view and find your Lucid documents with read-only permissions, removing any risk of data modification.

Plan availability: Free, Individual, Team, and Enterprise.
  • Not available to FedRAMP accounts.
  • Read the Lucid Plans article for more information about what is available on your account or the Upgrade your Lucid account article for instructions to upgrade.

 

This article is specific to the Lucid read-only MCP server. For information about the standard Lucid MCP server, check out our article.

Set up the Lucid read-only MCP server

The Lucid read-only MCP server is compatible with any MCP client, including those listed below.

While the setup process varies depending on the client you use, the server URL required to set up the Lucid read-only MCP server is the same:

  • https://mcp.lucid.app/mcp/readonly
    • This URL is not intended to be visited in your browser. It is a parameter you will use to connect the AI tool to Lucid’s MCP server.

The Lucid read-only MCP server can connect to any MCP client. Steps to connect vary between AI tools. Refer to documentation from your preferred AI tool for setup instructions.

Note: AI tools you set up with the Lucid read-only MCP server are Non-Lucid Applications (as defined in our Terms of Service). Your use of these AI tools is subject to the agreement between you and the third-party provider, including applicable licensing requirements, plan limitations, and version compatibility.

JSON config requirements

Some AI tools may require you to manually edit a JSON config file to connect them to an MCP server.

If your AI tool supports the HTTP transport method, use this JSON config:

"Lucid Software": {
"type": "http",
"url": "https://mcp.lucid.app/mcp/readonly"
}

If your AI tool does not support the HTTP transport method, you need to run a local server to forward HTTP requests to Lucid’s MCP server. This can be accomplished using the mcp-remote Node.js package. Make sure you have Node.js installed on your computer, then use the following JSON config:

"Lucid Software": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.lucid.app/mcp/readonly"]
}

Authenticate with an API key

The Lucid MCP server uses OAuth by default, which requires a user to approve access the first time an AI tool connects. For autonomous processes that run without a person available to grant that approval, you can authenticate using an API key instead. To set up authentication with an API key, follow these steps:

  1. Generate an API key in the Lucid developer portal.
  2. Configure your MCP client to use the API key for authentication.
  3. The MCP client will authenticate using the permissions granted to that key automatically, with no login or approval prompt required at runtime.

For admins: Enable or disable use of the Lucid read-only MCP server for your account

Admins on Team and Enterprise accounts can control whether or not to give users on their account access to the Lucid read-only MCP server.

To restrict or permit access to the Lucid read-only MCP server on your account:

  1. Select Admin from the navigation menu on the left side of the Lucid Home Page.
  2. Click Security from the navigation menu on the left side of the page.
  3. Select AI controls from the options that appear below "Security".
  4. Under “MCP access”, use the toggle next to "Allow users to connect”.
  5. Click Enable MCP or Disable MCP in the modal that appears.
  6. Click Save changes in the top-right corner.

Available actions with the Lucid read-only MCP server

Once the Lucid read-only MCP server has been set up, you can use it to complete various actions in Lucid by entering a prompt into your AI tool. The following use cases are currently supported:

Feature Example
Fetch content "Pull the text and structural content from the 'User Onboarding Flow' document so we can analyze it."
Fetch an image “Find the ‘Product Mockups’ document in Lucid and fetch the original source image attached to the ‘Feature A Screenshot’ shape.”
List threads on a document "Are there any open comment threads on the 'Marketing Campaign' document? List them for me."
List comments in a document thread “Show me all the individual replies and comments inside the 'Design Feedback' thread on that wireframe.”
Export document as PNG "Export the 'Sprint 4 Retrospective' board as a PNG so I can share it with the team."
List folder contents "Show me all the files and subfolders inside the 'UX Research' folder in Lucid."
Search within a document "Search the 'System Architecture' diagram for any mentions of 'database' or 'AWS'."
Search for documents "Find all Lucid documents related to the 'Q3 Product Roadmap' created in the last month."
Submit feedback "Submit a feedback log to Lucid saying that the document loading speed has significantly improved this week."

FAQ

What is the difference between the Lucid read-only MCP server and the standard Lucid MCP server?
The Lucid read-only MCP server is limited to viewing, searching, and analyzing your workspace, allowing your AI assistant to find files, read comment threads, search text, and export documents without risk of editing or deleting your data. In contrast, the standard Lucid MCP server includes both view and edit permissions.

How is my data used by the Lucid read-only MCP server?
The Lucid read-only MCP server acts as a pass-through and does not retain document content, user prompts, or queries. Your credentials are never shared outside your Lucid account. Usage of the Lucid read-only MCP server is subject to the negotiated services agreement between the parties or Lucid’s standard Terms of Service available at https://lucid.co/tos.

Resources

To learn more about this feature, explore the additional resources below. These resources are available in English only.

External resources
To learn more about MCP servers, check out this external documentation:

Give feedback on this article

Have feedback about this article? Tell us about your experience here.

Did you find what you were looking for?

Still have a question or want to share what you have learned? Visit our Community   to get help and collaborate with others.