Visily MCP
Brings Visily designs into AI coding agents, enabling design-to-code generation, board discovery, and theme token application.
README
Visily MCP
The Visily MCP server brings your Visily designs into your AI coding agent. Point it at a Visily board or screen and your agent can find it, turn it into application code, or pull its theme tokens — using your Visily design system as the source of truth.
[!NOTE] Access requirement. The Visily MCP works only on workspaces on a Pro or Business plan where you are the workspace creator or have an editor license. Free workspaces and viewer/guest roles are not supported. This is enforced on every request.
What you can do
- Turn a Visily design into code. Generate production-ready code from a Visily screen or board — whether the UI was AI-generated in Visily or hand-authored. The agent rebuilds semantic, responsive components (default stack: React + TypeScript + Tailwind v4 + shadcn/ui) instead of pasting a static export.
- Find a board or screen and get its link. Ask for a design by description ("the board I was working on for the 4.6 release") and the agent returns a link to open it — without leaving your editor.
- Apply a board's theme. Pull a board's design system as shadcn/Tailwind v4 tokens and apply it to another app's styling.
Install & connect
Authentication is OAuth — your client prompts you to sign in to Visily on first use. There are no API keys to manage.
The current Visily MCP server URL is:
https://app.visily.ai/api/mcp
Each platform's drop-in files live under plugins/<platform>/.
Claude Code
claude plugin marketplace add visily-app/mcp-plugins
claude plugin install visily@visily
Then reload Claude Code and run /mcp → select visily → complete the OAuth sign-in. Full
walkthrough: installation-guides/claude.md.
Cursor
Cursor 2.5+ packaged plugin — install from a local clone in the agent chat:
/add-plugin ./plugins/cursor
Reload Cursor; the MCP server appears under Settings → MCP (OAuth on first use) and the
Visily skill loads automatically. Details: plugins/cursor/README.md.
Codex
codex plugin marketplace add visily-app/mcp-plugins
codex plugin add visily@visily
Authorize via OAuth on first use. Full walkthrough:
installation-guides/codex.md.
ChatGPT
ChatGPT connects to remote MCP servers via Developer Mode custom connectors (Plus, Pro,
Business, or Enterprise). Enable Settings → Connectors → Advanced → Developer mode, add a
custom connector pointing at the server URL above, and complete OAuth. Then paste
plugins/chatgpt/AGENTS.md into a Project/GPT's instructions.
Steps: plugins/chatgpt/SETUP.md.
Antigravity
Copy plugins/antigravity/.agents/ into your project root and merge
plugins/antigravity/mcp_config.json into your global
MCP config. Details: installation-guides/antigravity.md.
Kiro
Copy plugins/kiro/ into your Kiro powers directory (it contains POWER.md,
per-intent steering/ workflows, and mcp.json). Merge mcp.json into your Kiro MCP config
and authorize via OAuth. The Power activates on Visily URLs and find/build/theme intents.
Lovable
Commit plugins/lovable/AGENTS.md to your repo root and add the
Visily MCP connector in the Lovable UI. Details:
plugins/lovable/SETUP.md.
v0 (Vercel)
v0 supports custom MCP servers natively. Add the server URL above via
Settings → MCP Connections (or Add MCP in the prompt form) and authorize via OAuth,
then commit plugins/v0/AGENTS.md to your repo root. Details:
plugins/v0/SETUP.md.
Other editors (Streamable HTTP)
Any client that supports Streamable-HTTP MCP can connect with this config:
{
"mcpServers": {
"visily": {
"url": "https://app.visily.ai/api/mcp"
}
}
}
This covers VS Code (Copilot), Windsurf, Cline, Gemini CLI, and similar tools. Check your client's docs for where MCP server config lives; OAuth runs on first use.
[!NOTE] Google AI Studio does not natively support MCP connectors yet (it only works through third-party browser bridges, which we don't recommend). For a Google-native option, use Firebase Studio, which supports MCP servers via
.idx/mcp.json— drop the sameurlconfig there. We'll add a first-class recipe when AI Studio ships native MCP support.
Prompting your agent
Give the agent a Visily link and a clear intent. The agent extracts the IDs from the URL — it doesn't open the page.
Examples
- "Find the Visily board I was working on for the 4.6 release and give me the link."
- "Build this Visily screen as React + Tailwind:
https://app.visily.ai/projects/abc/boards/456/elements/789" - "Implement this board in Next.js, reusing my
src/components/uicomponents." - "Apply my Visily board theme to this app's
globals.css."
A Visily URL looks like:
https://app.visily.ai/projects/{projectId}/boards/{boardId}/elements/{elementId}
Structure your Visily file for better code
The more intent your design carries, the better the generated code:
- Name your screens and key groups semantically in Visily — names flow into the export and into the link-matching for "find my board."
- Lean on Visily's theme / design system so the agent can apply real tokens instead of guessing colors and radii.
- Ask for one screen at a time for large boards — smaller selections produce more reliable results.
Tools
| Tool | Use |
|---|---|
get_design_context |
Primary. Returns designData + theme + screenshot metadata for one element. |
get_screenshot |
Fetch a screenshot for visual verification. |
get_board_theme |
Board design system as shadcn tokens (JSON / CSS / Tailwind). |
list_workspaces, list_projects, list_boards, list_elements |
Discovery / find-a-board. |
get_workspace, get_project, get_board, get_element, get_element_export |
Advanced — metadata and raw exports rarely needed for the three use cases; prefer the tools above. |
The server also ships prompts (e.g. find_visily_board, generate_code_from_element,
apply_board_theme) and resources (visily://docs/...) that your agent reads on demand.
Custom rules (optional)
Set project-level guidance so output stays consistent — like onboarding notes for a new teammate.
Cursor (.cursor/rules/visily-quality.mdc):
---
description: Visily code-generation rules
alwaysApply: false
---
- Treat Visily `designData` as a reference, not final code — rebuild semantically.
- Apply the Visily `theme` tokens; never hardcode hex, radii, or spacing.
- Reuse existing components in `src/components/ui` instead of duplicating.
- Verify against `get_screenshot` for 1:1 layout before finishing.
Claude Code (CLAUDE.md):
## Visily MCP rules
- Treat Visily `designData` as a reference, not final code — rebuild semantically.
- Apply the Visily `theme` tokens; never hardcode hex, radii, or spacing.
- Reuse existing components instead of duplicating; verify with `get_screenshot`.
How this repo is built
plugins/ is generated from a single shared source (shared/) by scripts/build.mjs. To
change the integration, edit shared/ and run npm run build — never edit plugins/ by
hand. See CONTRIBUTING.md.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.