OmniDebugLink

dev.omnidebuglinkPublished by omnidebuglink.devOfficial publisherv0.2.0Updated Oct 24, 2026

remote debug iOS/Android/Unity/Godot/Flutter/RN/Web on real-device.ui-tree/screenshots/taps,tests.

VerifiedStreamable HTTPWeb executableBrowser AutomationDeveloper Tools

Overview

AI-generated overview

Remote MCP server that lets an AI coding tool inspect and drive a running iOS, Android, Unity, Godot, Flutter, React Native or Web app on a real device.

What you can do
List connected devices and check their status, plus list and run tasks against them over MCP tools such as list_devices, run_task, device_status and list_tasks. Inspect the running app's UI tree and take screenshots. Drive the app with taps, swipes and typed input. Read logs and measure performance on real hardware. Works across networks through a single outbound WebSocket from the device, with no USB cable, shared network or emulator.
When to use it
Useful when an AI coding agent needs to debug or test a mobile, game or web app on real hardware, including devices behind NAT or on another network, and when you want automated UI interaction and screenshots without an emulator or cable.
Requirements
A remote endpoint reached over streamable HTTP with OAuth login in the browser. A device token pair created in the provider's console, and the matching client SDK added to the app (Unity, Android, Flutter, iOS/macOS, React Native, Godot or Web). The device needs an outbound WebSocket connection. No environment variables or headers are declared.
Before you install
OAuth login and a per-device token pair are required; one token pair equals one device seat. Tokens must never be embedded in a release build, where only a runtime-issued token via the conditional pattern is allowed. The server can tap, swipe and type in the running app, so it changes app state. App data and device interaction pass through the provider's relay. The console is free only during the public beta.

Installation

In SourceWeft

  1. Open OmniDebugLink in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/dev.omnidebuglink%2Fomnidebuglink/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "omnidebuglink": {
      "type": "http",
      "url": "https://api.omnidebuglink.dev/mcp"
    }
  }
}

README

OmniDebugLink

Remote debugging and automated testing on real devices — from anywhere.

Add a small client SDK to your app, connect it with a device token, and any AI coding tool (Claude Code, Cursor, Codex, Gemini CLI, …) can inspect and drive the running app over Model Context Protocol: dump the UI tree, take screenshots, tap, swipe, type, read logs, measure performance — no USB cable, no shared network, no emulator.

Your App ──wss──> OmniDebugLink relay ──MCP──> AI coding tool
  • One outbound WebSocket from the device — works across networks, behind NAT, on real hardware.
  • The AI side sees plain MCP tools: list_devices → run_task / device_status / list_tasks …
  • One token pair = one device seat. Tokens are per-device; never embed one in a release build — release use goes through the conditional pattern.

Client SDKs

PlatformInstallRepo
UnityPackage Manager → add from git URLomnidebuglink_unity
AndroidJitPack: com.github.omnidebuglink:omnidebuglink_android:<tag>omnidebuglink_android
Flutterpubspec git dependency (ref: tag)omnidebuglink_flutter
iOS / macOSSwift Package (iOS 15+ / macOS 12+)omnidebuglink_apple_swift
React Nativenpm i omnidebuglink/omnidebuglink_react_native#<tag>omnidebuglink_react_native
Godotcopy the addons/omni_debug_link/ folder (Godot ≥ 4.2)omnidebuglink_godot
WebCDN build via jsDelivr (@<tag>)omnidebuglink_web

Each SDK repo has the full install steps and task tables for its platform. Pin a version tag (#vX.Y.Z / ref: / @vX.Y.Z) in production hosts.

Quickstart

  1. Create a device token pair at the console (free during the public beta).

  2. Connect your app: add the SDK for your platform and start it with the token — freely in debug builds; in release builds only via the conditional pattern (runtime-issued token, never embedded).

  3. Add the MCP server to your AI tool (OAuth login in the browser):

    bash
    claude mcp add --transport http odl https://api.omnidebuglink.dev/mcp

    Works with any MCP client: Cursor, Codex CLI, Gemini CLI, …

  4. Ask the AI to list_devices and start driving the device.

Integrating an SDK into your app

sdk-integration.md is written to be handed to your AI coding tool verbatim — per-platform install and connect snippets, the rules that must not be broken, and a verification checklist. Ask the agent to read it and wire the SDK in.

Building a client for another platform?

Any platform with a WebSocket client can join — the wire protocol is a single JSON frame format:

→ Third-Party Client Development Guide — protocol spec, a portable reference implementation, known pitfalls, and a pre-deployment verification checklist. AI coding agents can implement a compliant client from this guide alone, without reading any SDK source.

Links

Source: README.md at commit 512c8c6

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

1
  1. v0.2.0LatestSep 16, 2026

Score

75% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.