A quick update on free credits

We’ve temporarily paused free credits for new accounts while we address recent misuse. Thank you for your patience and understanding.

logo
  • Environments
  • Enterprise
  • Pricing
TeamApr 10, 2026

Eigent Mobile MCP: Phone Automation for AI Agents

Tao SunTao Sun
·
Douglas LaiDouglas Lai
Eigent Mobile MCP — phone automation with AI agents
Automate Everything with
AI Workforce on Desktop
Download Eigent

Category: Productivity · Mobile · Integration


Mobile MCP brings iPhone and Android apps into the same multi-agent workflows you already run in Eigent. Instead of stopping at the desktop, your AI coworkers can open apps on a connected device, read the screen, tap through UIs, and pass results to tools like Slack—with task splitting and a Mobile Phone Agent handling the phone side. This page walks through a concrete mobile automation example: find a nearby restaurant on the device, then report to a Slack channel, all from one prompt.

Eigent can now control your phone.

With the Mobile MCP integration, Eigent's agents aren't limited to your desktop anymore. They can open apps on a connected iPhone or Android device, navigate the UI, extract information, and act on it — all as part of a single workflow you trigger with one prompt. No manual handoffs, no copy-pasting between screens.

This use case walks through exactly how that works, using a real example: asking Eigent to find a nearby restaurant and report back to a Slack channel — entirely through the phone.


1Connect Your Phone to Eigent

Before you write a single prompt, you need to add Mobile MCP as a tool inside Eigent.

Open Eigent and click Connectors in the top navigation bar. You'll see MCP & Tools in the left sidebar — click it, then select the Your own MCPs tab and click + Add MCP Server. Paste in the following configuration:

{
  "mcpServers": {
    "mobile-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@mobilenext/mobile-mcp"
      ]
    }
  }
}

Click Install. Once it appears in your Added external servers list, Eigent has access to your connected device — iPhone or Android — and is ready to use it as part of any workflow.

Mobile MCP supports iOS real devices, iOS Simulator, Android real devices, and Android Emulator. The tutorial video uses an Android device, but the same setup works on iOS.


2Write Your Prompt — Just Describe the Outcome

You don't need to tell Eigent which app to open or how to navigate it. Just describe what you want done. Here's the prompt used in this example:

Arrange a team dinner. All tasks must be completed using my mobile phone. For a group of five people, find a nearby restaurant rated 4.5 or higher, then send the restaurant name, address, and rating details to the #team-dinner Slack channel.

That's it. One prompt spanning two devices and two apps.


3Eigent Splits the Task Automatically

Eigent doesn't execute your prompt as one big instruction. It first breaks it down through Task Splitting — identifying the distinct sub-tasks and which agents should handle them.

In this case it recognises two things need to happen: find a restaurant using the phone, then report the result via Slack on the desktop. It assigns the first to the Mobile Phone Agent and coordinates the handoff automatically. You don't configure any of this.


4The Mobile Agent Navigates Your Phone

The Mobile Phone Agent takes control of your connected device using a vision-based approach — it sees the screen the same way you would, and interacts with it through taps and navigation rather than relying on APIs. This works on both iOS and Android.

Here's what it does on the phone:

  • Opens Google Maps
  • Confirms the current location
  • Searches for nearby restaurants and filters for those rated 4.5 or higher
  • Scrolls through the results and selects a qualifying option
  • Taps into the listing to collect the name, address, rating, and other details

The agent is reading and interacting with the real Maps interface — live data, real filters, the same screen you'd use yourself.


5Eigent Reports to Slack

Once the mobile agent has what it needs, Eigent switches context to the desktop without any input from you. It opens Slack, navigates to the correct channel, and posts a formatted message with everything it found — restaurant name, star rating, address, and a Maps link.

The whole cross-device handoff happens inside a single workflow run.


6Review the Completion Report

When the task finishes, Eigent's dashboard displays a Completion Report summarising what happened:

  • Location confirmed on the mobile device
  • Restaurant found matching the criteria
  • Details posted to the Slack channel

This gives you a clear record of every action the agents took, so you can verify the result or adjust the prompt and rerun it.


7Why This Matters

Most agent workflows stop at the desktop. The moment a task requires something on your phone, you're back to doing it yourself.

Mobile MCP removes that boundary. Eigent can now operate any iOS or Android app — not just the ones with a developer API — as part of a larger cross-device workflow. Combined with Eigent's Multi-Agent coordination, this means a single prompt can reach across your desktop, your browser, your integrations, and now your phone.

The team dinner example is deliberately simple. The same capability applies anywhere you'd normally have to pick up your phone mid-workflow.


8What to Try Next

Once Mobile MCP is connected, try prompts that span your phone and your desktop tools:

  • Search for a venue on Google Maps and add it directly to Google Calendar.
  • Check a mobile-only app for information and save the output to a Notion page.
  • Use your phone to look something up and have the result summarised in Slack.
  • Run any workflow where the final step is a notification sent to your phone.
  • Combine phone-based research with a document generated by Eigent's Document Agent.

9Tips for Better Results

Specify "use my mobile phone" in the prompt. When a task could be handled by either the browser agent or the mobile agent, telling Eigent which to use makes the routing unambiguous.

Test the connection with a simple task first. Before running a complex workflow, try something like "open Google Maps on my phone and confirm my current location." If the agent can do that, the integration is working.

Pair mobile steps with desktop integrations. Mobile MCP works best as one part of a larger workflow. Combine it with Slack, Notion, Google Calendar, or any other tool already connected to Eigent to build end-to-end automations that move information across all your devices automatically.

Other use cases

Gemini 4: Vibe-Code a 3D Cinematic in Blender from One Prompt

Gemini 4: Vibe-Code a 3D Cinematic in Blender from One Prompt

Create a twelve-second, four-shot cinematic in the Cozy Kitchen scene: a draught lifts a recipe page off its stand, carries it across the work area, and leaves it draped on the strawberry tartlet. Preserve the warm illustrated kitchen. Add a thin two-sided RecipePage with readable recipe marks, follow the flight corridor in action.json, and show the calm setup, motivated lift, flight, contact, and settling with visible wind clues. Use four distinct cameras with camera-bound cuts, depth of field in shots 2 and 3, and smooth, measurable page and camera motion. Hand off a packed, editable sequence.blend, a shot list CSV, a labelled 3×3 storyboard, and nine stills.

Gemini 4: Turn Video Tutorials into Production RL Engineering Skills

Gemini 4: Turn Video Tutorials into Production RL Engineering Skills

Watch the listed YouTube talks and turn their methods and insights into 2–4 reusable, file-based agent skills, using only these videos as primary sources. For each talk, build a timestamped outline, record the key insights, results, and warnings, and label every claim OBSERVED, INFERRED, or UPDATED. Give each skill a SKILL.md with a key insights section and a tool-agnostic path, plus an evidence.md. Validate every skill, then publish a light-mode HTML report served from 127.0.0.1 with the talks embedded, clickable timestamps, a comparison of the talks, a coverage matrix that shows gaps, and a clear list of limitations.

Build a CAD Model from Video: Gemini 3.7 Flash vs. 3.6 Flash

Build a CAD Model from Video: Gemini 3.7 Flash vs. 3.6 Flash

Analyze the transformers character in the uploaded reference video and generate a production-ready, highly detailed 3D CAD/mesh model exported as a .glb file. Analyze the character identity and style, joints and articulation, and surface details. Create ultra-high-precision hard-surface mechanical geometry with distinct body and armor pieces separated around logical joint pivot points. Use metallic PBR materials, including anodized metals, brushed steel, dark titanium, high-gloss painted armor, and emissive channels for the eyes, chest core, and joint glow effects. Pack the diffuse, normal, roughness, metallic, and emissive textures directly into the GLB. Ensure clean topology, manifold geometry, and real-time optimization without losing mechanical sharpness.

Automate everything with AI workforce on desktop
Download Eigent

Try Eigent today

Download the open-source desktop app. Your AI workforce, running on your machine.

Download Eigent
Eigent

Get the latest updates, tutorials, and releases on AI workforce automation.

Thank you for subscribing!

ProductEigentEnvironmentsPricingEnterprise
ExploreSolutionsUse CasesSkillsPluginsBlogs
DevelopersDocsGitHubCAMEL-AIOpen Source FundPartner
DownloadFor open source
CompanyAbout UsBrandCareersSupportTerms of UsePrivacy PolicySecurity & TrustCookie PolicyRefund & Trial Policy

All rights reserved © 2026 EIGENT UK LTD