onUI

What is onUI?

onUI is a free, open-source browser extension that turns visual feedback on a web page into something an AI coding agent can act on. In Annotate mode you click an element, attach an intent and a severity, and use Shift to select several components at once; in Draw mode you drag rectangles or ellipses over spacing, alignment and layout problems. The notes can be exported in four detail levels (compact, standard, detailed and forensic), or read straight from the page by an agent through the local onui-local MCP server using tools such as onui_get_report and onui_search_annotations. That sets up a loop where you annotate, the agent fixes the code, and you review and annotate again. onUI needs no changes to your app and has no backend, since the extension, native host and MCP server all run on your machine. It is enabled per tab, works in Chrome, Edge and Firefox, and is described as working with Claude Code and Codex. There is no price on the site; it is distributed as an open-source project.

Use cases

• Annotate UI bugs for an AI agent
• Mark layout and spacing problems with drawn regions
• Export UI feedback at several detail levels
• Run an annotate-and-fix loop with Claude Code
• Flag visual issues with intent and severity
• Review front-end changes without editing app code

Who it's for

• Front-end developers
• UI designers
• QA engineers
• AI-assisted developers

Why use onUI

• Gives an AI agent precise element-level context instead of a vague description of what looks wrong.
• Keeps all annotations on your device, since the extension, native host and MCP server run locally.
• Works on any page where content scripts can run, with no code changes or instrumentation in your app.
• Lets you choose how much detail to send, from a compact summary to a forensic report.
• Supports a repeatable loop where the agent reads fresh annotations, fixes the UI and you re-check.

onUI features

• Annotate mode with intent and severity per element
• Shift multi-select for issues spanning several components
• Draw mode with rectangle and ellipse regions
• Four export formats: compact, standard, detailed, forensic
• Local onui-local MCP server with six tools
• Agent reads annotations via onui_get_report and onui_search_annotations
• Per-tab enable, off by default on new tabs
• No app instrumentation, backend or cloud dependency
• Keyboard shortcuts for toggling Annotate and Draw modes
• Annotate mode freezes the page so element positions stay stable

onUI integrations

• Claude Code
• Codex
• MCP

How to get started with onUI

1. Install onUI from the Chrome Web Store or Edge Add-ons, or run the release installer script, which can also register the onui-local MCP bridge.
2. Open the page you want to review and enable onUI for that tab.
3. Use Annotate mode to click elements and add intent and severity, or Draw mode to mark layout regions.
4. Copy the export in the detail level you need, or let your agent read the annotations from the local MCP tools.
5. Have the agent apply fixes, then review the page and add new annotations until the UI looks right.

onUI limitations

• Annotate mode freezes the page, so you must exit it to scroll
• Firefox setup uses a temporary add-on loaded from the generated manifest
• The extension is tab-scoped, so each new tab starts with onUI off
• Store versions can lag the latest release installer

onUI pricing

onUI is open source, so you can use and self-host it for free.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

onUI alternatives

The main onUI alternatives listed on Debuly are Applitools, BridgeMind, Continue, Graft (now Trail-code), Kilo Code and NestMux. The table compares pricing and platforms; order is A–Z.

1. 

   Applitools

   Visual AI testing platform for catching UI and functional bugs across browsers and devices, with a 14-day free trial.

   CodingDesignAutomation and agents
   Free trialVisit Applitools (official site)
2. BM

   BridgeMind

   Desktop app for running AI coding agents, terminals and chat threads in one window, with voice dictation and a plugin catalog.

   CodingAutomation and agentsDeveloper platforms
   PaidVisit BridgeMind (official site)
3. 

   Continue

   Open-source AI coding assistant for VS Code and JetBrains with chat, agent, edit and autocomplete modes, now part of Cursor.

   CodingDeveloper platforms
   Open sourceVisit Continue (official site)
4. 

   Graft (now Trail-code)

   Open source context layer that maps a codebase for coding agents and stores team learnings as Markdown files in git.

   CodingDeveloper platformsAutomation and agents
   Open sourceVisit Graft (now Trail-code) (official site)
5. 

   Kilo Code

   Open-source AI coding agent for VS Code, JetBrains and the CLI, with 500+ models, bring-your-own-key and team plans.

   CodingDeveloper platformsAutomation and agents
   FreemiumVisit Kilo Code (official site)
6. 

   NestMux

   Desktop terminal workspace that runs Claude, Gemini, Codex and Copilot agents side by side, with git worktrees and team sharing.

   CodingDeveloper platformsProductivity
   FreemiumVisit NestMux (official site)

onUI FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from onui.onllm.dev · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.