Beta feature: The Frontify MCP server is currently in beta. During this experimental phase, it may be incomplete, unstable, or subject to breaking changes. No support commitments or reliability guarantees are provided. See Things to know before you start for details.
We'd love to hear what works and what doesn't. Share your feedback here.
The Frontify MCP server connects Frontify to compatible AI tools using the Model Context Protocol (MCP). Once connected, an AI agent can access brand guidelines, assets and templates directly from Frontify — without leaving the tool where work is happening.
This means teams using AI tools like Claude, ChatGPT, Cursor, or Figma can produce brand-aware outputs grounded in the same structured brand knowledge that lives in Frontify. No custom integration is required.
What is MCP
Model Context Protocol (MCP) is an open standard that lets AI models interact with external tools and systems in a structured way. It works like a translator between an AI agent and an API, turning natural language prompts into precise actions.
In practice, the Frontify MCP server wraps the Frontify API into a set of tools that AI agents can understand and use. Instead of building custom integrations, teams connect once and start working.
What you can do with it
The Frontify MCP server provides 52 tools organized across several areas of functionality. Here are some common tasks it enables:
Find and access brand assets. Search across brands, libraries, and projects to locate logos, images, videos, and documents directly from a connected AI tool.
Read brand guidelines. Surface tone of voice rules, style guides, and documentation pages without opening Frontify.
Generate on-brand creative. Use brand templates to produce finished designs with custom text, imagery, and layout — all from a single prompt.
Organize and tag assets. Upload files, move them into the right folders, add tags, assign metadata, and build collections.
Add alt text in bulk. Point an AI agent at a folder and let it generate meaningful alternative text for each asset based on existing metadata.
Route assets for review. Create workflow tasks, add checklist items, and assign assets to the right approval stage.
Extend metadata schemas. Create new custom metadata fields without navigating the admin interface.
Tool packs
The Frontify MCP server groups its tools into packs — bundles of functionality tailored to different use cases and risk levels. There are two ways to use packs:
Connect using the main URL (
https://mcp.frontify-integrations.com/mcp). This starts with the default pack and lets administrators switch between packs at any time from within the AI tool. Only one pack can be active at a time, and switching affects all users on the account.Connect using a pack-specific URL (e.g.,
https://mcp.frontify-integrations.com/mcp/packs/discovery). This locks the connection to a single pack. Pack-specific URLs are available at any time, regardless of what an administrator has set on the main URL.
The default pack is discovery, which provides read-only access to brands, guidelines, libraries, projects, assets, and templates. It includes 25 tools and does not make any changes to Frontify data, making it the safest starting point.
Additional packs extend functionality with write actions such as uploading and organizing assets, managing workflows, and generating creative output from templates. Each pack is scoped to a specific set of capabilities so that organizations can match their level of access to their comfort with AI-assisted actions.
Explore all available packs, their tools, and their URLs at mcp.frontify-integrations.com.
Things to know before you start
The MCP server is in beta. During this experimental phase, there are a few important considerations:
Write actions can change Frontify data. Packs that include write tools can create, modify, or move assets, metadata, and workflow tasks. AI agents may occasionally misinterpret instructions, which could lead to unintended changes. Start with the read-only discovery pack and back up important data before enabling write-capable packs.
AI output should always be reviewed. AI models can produce results that look correct but aren't. Always verify outputs — especially before publishing, sharing, or acting on them.
Data leaves Frontify when sent to an external AI tool. The MCP server gives AI agents structured access to Frontify data, but once that data reaches the external tool (for example, Claude or ChatGPT), it is processed within that vendor's infrastructure. Frontify does not control how external AI vendors handle, store, or retain data. Organizations with strict data residency or compliance requirements should review their AI vendor's data policies before connecting.
No formal support is provided during beta. In this experimental phase, the MCP server may be incomplete, unstable, or subject to breaking changes at any time.
Pricing
The Frontify MCP server is free to use during beta. There is no additional cost, and usage does not count toward monthly active users (MAU). Pricing may be introduced once the feature moves beyond beta. If that happens, you will be notified in advance. There will be no surprise charges.
How to activate the MCP server
The MCP server is not enabled by default. To activate it:
Contact your customer success manager (CSM) or Frontify Support and request that the Frontify MCP be enabled for your account.
Once enabled, all tool packs are available using their specific URLs.
Connect to Claude
Users can connect Frontify to Claude, Claude Desktop, or Cowork using a custom connector. Custom connectors are available on Claude Free, Pro, Max, Team, and Enterprise plans. Free plan users are limited to one custom connector.
To connect:
Follow the instructions in the Claude custom connectors guide to create a new connector.
When prompted for a URL, use the URL for the tool pack you want to use. If you haven't decided on a pack yet, use the main URL (
https://mcp.frontify-integrations.com/mcp) for the most flexibility — this allows switching packs at any time without updating the connector.
Connect to Figma Make
Users can connect Frontify to Figma Make using a custom MCP connector. A custom connector is visible only to its creator. On Figma Organization and Enterprise plans, administrators can publish custom connectors to the organization.
To connect:
Follow the instructions in the Figma custom MCP connector guide to create a new connector.
When prompted for a URL, use the URL for the tool pack you want to use. If you haven't decided on a pack yet, use the main URL (
https://mcp.frontify-integrations.com/mcp) for the most flexibility.
Connect using a configuration file
Some AI tools (such as Cursor, Windsurf, or Claude Desktop) support MCP connections through a JSON configuration file. Add the following to the mcpServers section of the configuration:
{ "mcpServers": { "frontify": { "url": "https://mcp.frontify-integrations.com/mcp/packs/discovery" } } }Replace the URL with the pack URL that matches the desired level of functionality. Pack URLs are available at mcp.frontify-integrations.com.
Switch tool packs
There are two ways to switch the active tool pack:
At runtime: If the account is connected using the main URL (
https://mcp.frontify-integrations.com/mcp), administrators can switch packs from within the AI tool by using thefrontify_account_configure_packtool. The current configuration can be checked usingfrontify_account_get_config.By changing the URL: Update the connector or configuration file to point to the URL of a different pack. Each pack has a dedicated URL available on mcp.frontify-integrations.com.
Share feedback
The MCP server is in beta and under active development. No formal support is available during this phase. To share what's working, report issues, or suggest improvements, use the feedback form.
More articles about the MCP server and other experiments are published on Frontify Kitchen.