Allows capturing and annotating UI regions in Firefox to provide visual context for coding agents.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@uigrep pull the latest UI feedback from the browser for the demo page"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
uigrep is a local-first visual context bridge for coding agents. Start the background app, press a shortcut, annotate one or more UI regions in the browser, and let an MCP-enabled coding harness retrieve only the context it needs.
The desktop UI is a fixed, non-expandable status pill.
Capture happens in the user's real browser.
The harness current working directory is the codebase root.
Context is stored locally and disclosed progressively.
No cloud account, telemetry, or hidden model calls.
Related MCP server: markupR MCP Server
apps/desktop — Tauri desktop shell, local daemon, and status pill.
apps/extension — WXT browser companion for Chromium and Firefox.
apps/test-app — deterministic UI used by integration tests.
packages/schema — canonical capture contracts and budgets.
packages/bridge — authenticated loopback protocol helpers.
packages/mcp — generic MCP stdio adapter.
Node.js 22+
pnpm 10+
Rust stable and the platform prerequisites for Tauri 2
Run pnpm install.
Run pnpm check.
Run pnpm --filter @uigrep/desktop tauri dev for the desktop app.
Run pnpm --filter @uigrep/extension dev for Chromium or pnpm --filter @uigrep/extension dev:firefox for Firefox.
Run pnpm --filter @uigrep/test-app dev for the fixture page.
The initial default shortcut is Alt+Shift+G (Option+Shift+G on macOS), and it will be configurable.
This server cannot be deployed
Give AI coding agents access to your Vynix visual feedback, bug reports, and AI diagnosis.
Comment on AI-generated webpages; feedback flows back to your coding agent. Free, MIT, local-first.
Browser-backed QA with evidence and fix-ready reports for coding agents.
Human feedback for AI agents: share HTML, get a live review link, read anchored notes as markdown.
