onUI

onUI

Annotate elements, draw regions, and ship cleaner UI faster.

Category
Visit Server

README

onUI

Annotate Any UI for AI Agents

Lightweight browser extension (Chrome + Edge + Firefox) + local MCP bridge for annotation-first UI pair programming.

Powered by onLLM.dev.

GitHub stars GitHub license Chrome Stable Edge Stable Firefox Stable Buy Me A Coffee

[!NOTE] onUI is now stable and production-ready.

Demo

<a href="https://github.com/onllm-dev/onUI/raw/main/landing-page/assets/onui-demo.mp4"> <img src="https://github.com/onllm-dev/onUI/raw/main/landing-page/assets/onui-demo-preview.gif" alt="onUI demo โ€” click to play full video" width="720"> </a>

<sub>Click the preview above to play the full demo video.</sub>

โœจ Why onUI

  • ๐Ÿงฉ No integration into app code
  • ๐ŸŽ›๏ธ Per-tab ON/OFF control (off by default)
  • ๐ŸŽฏ In-page annotation dialog with intent + severity
  • โœ๏ธ Draw mode for region annotations (rectangle + ellipse)
  • โš™๏ธ Compact toolbar with pop-out settings (output level + clear on copy)
  • ๐Ÿ‘€ Visual markers and hover targeting
  • ๐Ÿงพ Export outputs in compact / standard / detailed / forensic formats
  • ๐Ÿ›ก๏ธ Shadow DOM isolation for stable styling
  • ๐Ÿ”Œ Local MCP server + native bridge (no cloud backend required)

Install (Current)

Option A: Browser Extension Stores (recommended)

Chrome Web Store: https://chromewebstore.google.com/detail/onui/hllgijkdhegkpooopdhbfdjialkhlkan?authuser=0&hl=en-GB

Microsoft Edge Add-ons: https://microsoftedge.microsoft.com/addons/detail/onui/fkcmlckehjhcicihbnmhkadfhjhfnond

Option B: One-command installer from latest GitHub release

Use this if you want the latest version (store updates may lag) or for Firefox:

macOS/Linux:

curl -fsSL https://github.com/onllm-dev/onUI/releases/latest/download/install.sh | bash

macOS/Linux (Firefox artifact):

curl -fsSL https://github.com/onllm-dev/onUI/releases/latest/download/install.sh | bash -s -- --firefox

Windows (PowerShell):

irm https://github.com/onllm-dev/onUI/releases/latest/download/install.ps1 | iex

Windows (PowerShell, Firefox artifact):

iwr https://github.com/onllm-dev/onUI/releases/latest/download/install.ps1 -OutFile install.ps1; .\install.ps1 -Firefox

The installer handles extension install and can set up MCP in the same run. When prompted with Set up local MCP bridge now? [y/N], enter y to enable MCP.

Then load it in Chrome or Edge:

  1. Open chrome://extensions or edge://extensions
  2. Enable Developer mode
  3. Click Load unpacked
  4. Select ~/.onui/extensions/current (or %USERPROFILE%\.onui\extensions\current on Windows)

Chromium browsers require this final manual step for unpacked extensions.

Firefox (manual from release artifact):

  1. Download onui-firefox-add-ons-vX.Y.Z.zip from the GitHub release page.
  2. Extract it to a local folder.
  3. Open about:debugging#/runtime/this-firefox.
  4. Click Load Temporary Add-on...
  5. Select the extracted manifest.json.

๐Ÿง  Usage

  1. Open any supported website tab.
  2. Click the onUI extension icon.
  3. Enable This Tab.
  4. Use the on-page launcher to open the compact toolbar.
  5. Toggle Annotate mode for element targeting or Draw mode for region targeting.
  6. Hold Shift and click multiple elements to batch-select targets.
  7. Release Shift to open a shared annotation dialog for selected targets.
  8. Save once to create one annotation per selected element (or one region annotation in draw flow).
  9. Open toolbar Settings to choose output level and configure Clear on copy.
  10. Copy exported output from the toolbar.

๐Ÿ”Œ Local MCP Setup

onllm-dev/onUI MCP server

Recommended path: use the same installer command above and answer y when prompted.

If you want to force MCP setup in non-interactive mode:

macOS/Linux (--mcp):

curl -fsSL https://github.com/onllm-dev/onUI/releases/latest/download/install.sh | bash -s -- --mcp

Windows (PowerShell, set env var before running installer):

$env:ONUI_INSTALL_MCP=1; irm https://github.com/onllm-dev/onUI/releases/latest/download/install.ps1 | iex

MCP setup now uses a prebuilt release bundle (no local build required), but still needs Node 20+.

Manual MCP setup from source is still supported:

pnpm build:mcp
pnpm setup:mcp
pnpm doctor:mcp

Manual JSON config for custom MCP routers/clients

If your MCP router uses an object-style mcpServers map, use this canonical entry:

{
  "mcpServers": {
    "onui-local": {
      "command": "node",
      "args": [
        "/ABSOLUTE/PATH/TO/onUI/packages/mcp-server/dist/bin/onui-cli.js",
        "mcp"
      ]
    }
  }
}

Use an absolute path for onui-cli.js (relative paths are often rejected or resolved incorrectly by routers).

If your router uses a list/array schema instead of an object map, adapt the same command/args shape like this:

{
  "servers": [
    {
      "name": "onui-local",
      "command": "node",
      "args": [
        "/ABSOLUTE/PATH/TO/onUI/packages/mcp-server/dist/bin/onui-cli.js",
        "mcp"
      ]
    }
  ]
}

The list example above is a schema adaptation pattern, not a claim about any specific router's exact key names.

Setup/verification notes:

  • Run pnpm build:mcp first so packages/mcp-server/dist/bin/onui-cli.js exists.

  • Keep the server entry name as onui-local.

  • Run pnpm doctor:mcp after wiring config to confirm local setup health.

  • Auto-registers onui-local for Claude Code and Codex when those CLIs are installed.

  • Browser support in this release: Chrome stable + Edge stable + Firefox stable (unpacked).

  • @onui/mcp-server is workspace-local (private: true), so run setup/doctor from this repo.

See:

  • docs/mcp-setup.md
  • docs/doctor.md
  • docs/release.md

Maintainer Build + Release

app.sh is the local release entrypoint (no CI/CD dependency).

Local validation + artifact packaging

./app.sh --build

This runs:

  1. Prereq checks (Node 20+, pnpm, git, zip)
  2. Build order: @onui/core -> @onui/extension -> @onui/mcp-server
  3. MCP tests
  4. MCP doctor smoke check (warnings allowed, errors fail)
  5. Artifact packaging into artifacts/vX.Y.Z/

Artifacts:

  1. onui-extension-unpacked-vX.Y.Z.zip
  2. onui-chrome-web-store-vX.Y.Z.zip (manifest key stripped for CWS)
  3. onui-edge-add-ons-vX.Y.Z.zip (manifest key stripped for Edge Add-ons)
  4. onui-firefox-add-ons-vX.Y.Z.zip
  5. onui-mcp-bundle-vX.Y.Z.zip
  6. install.sh
  7. install.ps1
  8. checksums.txt

Local release + GitHub publish

./app.sh --release

Release gates:

  1. Clean git tree
  2. Current branch is main
  3. gh auth status succeeds

Release actions:

  1. Auto patch bump from root package.json
  2. Sync version across extension + MCP runtime strings
  3. Run full --build
  4. Commit + tag vX.Y.Z
  5. Push commit/tag
  6. Create GitHub release with packaged assets

๐Ÿ› ๏ธ Development

pnpm install
pnpm check
pnpm test:coverage

๐Ÿ—‚๏ธ Repository Structure

packages/
  core/        Shared annotation/report types + formatters
  extension/   Browser extension runtime (background/content/popup)
  mcp-server/  Local MCP server + native bridge setup/doctor tooling

โญ Support

If onUI is useful to you, please star the repo: https://github.com/onllm-dev/onUI

It helps other users discover the product.

<a href="https://buymeacoffee.com/tushar_s" target="_blank"><img src="https://cdn.buymeacoffee.com/buttons/v2/default-yellow.png" alt="Buy Me A Coffee" height="48"></a>

Star History

Star History Chart

๐Ÿ“„ License

GPL-3.0

Recommended Servers

playwright-mcp

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.

Official
Featured
TypeScript
Magic Component Platform (MCP)

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.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

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.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

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.

Official
Featured
TypeScript
Kagi MCP Server

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.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

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.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured