Nila MCP Server
Nila's MCP server connects AI assistants directly to your Nila workspace. It uses the Model Context Protocol, an open standard for connecting AI clients to external tools and data. This lets MCP-capable clients read and work with your Nila design files.
It's a workspace-level integration rather than a read-only bridge. Broadly, what you can do through the Nila editor, an assistant can do through MCP.
Designs are built using Nila's own primitives, including shapes, text, groups, and screens. The result is a native Nila file that you can open and continue editing, rather than an export or a rebuilt copy.
What you can do with it
Design with your design system. An assistant can use real components from your team and organisation libraries, set their properties and variants, bind values to your variables, and apply shared styles. This keeps generated designs aligned with the system your team already uses.
Build and maintain your design system. Create and organise components, variables, collections, and modes, manage styles, and publish a library when it's ready.
Keep projects in sync. Pull library updates into a project as your design system changes, and find work that has drifted from the system, such as detached styles, off-system colours, or components that no longer match their masters.
Design from a prompt. Ask for a settings screen, pricing page, or other design and have it built directly on a Nila page using the components and styles your team already has.
Work on what already exists. Point an assistant at an existing screen and have it restyle, re-layout, rewrite the copy, or extend it to support more states.
Move work between design and code. Read layout, styles, variables, and component structure from a page as structured data that your tooling can consume.
What the assistant can reach
| Area | What it covers |
| Components | Browse team and organisation libraries, place instances, set variants and properties |
| Variables and styles | Variable collections, modes and values; shared colour, text and effect styles |
| Library lifecycle | Adopt assets into a project, pull library updates, publish a library |
| Workspace | Editions, teams, projects and pages; recent work |
| Projects and pages | Create, rename, reorder, trash and restore |
| Reading a design | Page contents, screens, individual objects, and batch reads |
| Creating and editing | Shapes, text, screens and components; vectors, freehand strokes and SVG import |
| Layout and structure | Spatial queries, hierarchy traversal, relationships, anchoring and auto layout |
| Text | Text content, paragraphs and character styling |
| Images and guidelines | Upload images to a page, place layout guidelines |
Where you can use it
Any AI client that can connect to an MCP server can use it, so you're not tied to a particular editor or assistant.
In practice, that includes three types of clients:
Coding agents and IDEs. Claude Code, Codex, Cursor, VS Code, and Xcode. This is useful for design-to-code workflows, where an assistant can read a canvas's structure, variables, and component usage and write code against the actual design instead of a screenshot.
Chat assistants. Claude desktop and web, along with other MCP-capable chat clients. These are useful for working directly with designs, such as creating a screen, restyling an existing one, or checking a file against your design system.
Your own tooling. You can also connect any MCP client library to build custom workflows, including scripted or batch operations across multiple files.
The server is hosted by Nila at https://nila.zoho.in/mcp, so there's nothing to install or keep running locally. Add the URL to your client, sign in with your Nila account in the browser when prompted, and your client can connect to your workspace.
You don't need to create or manage API keys or tokens.
Support for remote MCP servers varies between clients and continues to change, so check your client's documentation for the latest support.
Found this helpful?
Can't find what you need?
Try refining your search or explore our Help Center