Try free

Last updated 13 October 2026

View Claude Code sessions on a phone

Managing long running coding tasks

You are in the middle of a massive code migration. You have initialized Claude Code in your repository and started the agent using the command claude. The agent is currently using the Read tool to examine your existing structure and the Glob tool to find all relevant files for the update. You want to see the progress of the Bash tool as it runs your test suite, but you need to leave your desk to handle other tasks. When you are away from your screen, you lose the ability to see if the agent has hit a roadblock or if the build has failed.

Traditional terminal setups do not allow you to easily monitor these sessions from a distance. You might try to use a remote desktop app, but those are often clunky and hard to read on a small phone screen. If the agent enters a state where it is waiting for your input, the whole process sits idle until you return. This creates a bottleneck in your development workflow. You need a way to keep an eye on the terminal output and interact with the agent without being physically tied to your computer chair for hours at a time.

Tracking tool execution and progress

Claude Code uses a variety of tools to perform its work. It might use the Grep tool to search for specific strings across your project or the Edit tool to modify files. Each of these actions appears in the terminal with a specific status. If you are running a complex task, the agent might generate a long series of tool calls. Understanding where the agent is in its plan requires looking at the terminal history. You can use the up and down keys or Ctrl+R to look through your own shell history, but the agent's internal progress is visible only in the active session window.

The agent also provides a visual representation of its work using symbols and text colors. If it is running the Bash tool to execute a long script, you will see the output stream in real time. If the script fails, the agent will analyze the error and try to fix it. This autonomous behavior is powerful, but it still requires human oversight to ensure the agent does not go down a wrong path. Monitoring this output from a mobile device allows you to catch these mistakes early, even if you are in another room.

Organizing multiple project sessions

Modern development often involves working on multiple repositories or different branches of the same project simultaneously. You might have one instance of the agent fixing a CSS bug in a frontend folder while another instance is refactoring a database schema in the backend. You can use the command claude --resume to see a list of your previous sessions and pick one to continue. However, managing five or six open terminal tabs becomes mentally taxing as you try to remember which tab is doing what.

Each session has its own context and history. If you use the command claude --continue, you carry on with the most recent session in that specific folder. This is helpful for staying focused, but it does not give you a high-level view of all your active work. You need a system that organizes these sessions into a clear layout so you can monitor the status of every project at once. This visibility is especially important when you are monitoring several long-running builds from your phone.

Visualizing sessions with a dedicated interface

A native application can provide the structure that a standard terminal lacks. adhdTerminal is a native app for macOS and Windows that runs many Claude Code sessions side by side. Each session is displayed as a tile on a wall, and every tile has its own color, number, and name. This allows you to categorize your work visually. You might use blue for frontend tasks and green for backend tasks. The interface shows whether a tile is working, waiting for your input, done, or idle.

This tiling system makes it easy to see the last exchanges of every conversation without clicking through different tabs. The tiles stay in the same place even after you restart your computer or the app. Because adhdTerminal is made by MINDBALANCER OU in Estonia and is not affiliated with Anthropic, it acts as a specialized wrapper designed for developers who need to manage high cognitive loads. It costs 99 dollars once and provides a 7 day free trial without requiring an account.

Connecting the mobile mirror app

The true benefit of a tiled terminal wall is the ability to take it with you. There is a free iPhone app that mirrors the adhdTerminal wall over an end to end encrypted link. This ensures that your private code and session data remain secure while being transmitted to your mobile device. The phone app shows the same colors, numbers, and names as your desktop wall. You can see the status of every tile at a glance while you are away from your desk.

The mobile app is not just a passive viewer. You can select a tile to see the full conversation and the most recent tool outputs. If the agent asks a question or hits a point where it needs a new task, you can type your response directly into your phone. This input is sent back to your computer and injected into the active session. This allows you to keep the agent productive during your commute or while you are away from your workstation for a break.

Handling permissions and user prompts

Claude Code operates in different permission modes that determine how much autonomy it has. The default mode requires you to approve most tool uses, especially when the agent wants to use the Edit or Write tools. You can cycle through these modes by pressing Shift+Tab inside a session. If you are in plan mode, the agent only suggests changes without making them. If you are in bypassPermissions mode, it will act without asking you first.

When you are monitoring your work from a phone, the permission settings in .claude/settings.json become very important. You can configure the agent to allow certain tools while requiring approval for others. If the agent reaches a point where it needs permission to proceed, the tile on your wall and the mirror on your phone will show that it is waiting for you. You can then review the proposed changes on your mobile screen and grant permission with a single prompt, allowing the work to continue uninterrupted.

Configuring project files for better feedback

To make the mobile experience more effective, you should use the CLAUDE.md file. This file sits in your repo root and holds project instructions that the agent reads at the start of every session. You can use this file to tell the agent exactly how you want it to report its progress. For example, you can instruct it to provide a summary of its accomplishments after every five successful tool uses. This makes the text on your phone screen much easier to digest.

You can also use the .claude/settings.json file to customize how the agent behaves during long tasks. By setting specific hook events like PostToolUse or Notification, you can create a workflow that highlights the most important information. If you have a CLAUDE.md file in a subdirectory, those instructions will apply to that specific part of the tree. This hierarchical approach ensures that the agent always has the right context, which reduces the number of clarifying questions it needs to ask you while you are mobile.

Controlling cost and session health

Running multiple AI sessions can lead to significant costs if you are not careful. You can use the /cost and /usage slash commands at any time to see how much you have spent. Monitoring these figures from your phone helps you catch sessions that might be stuck in a loop or performing unnecessary work. If you see the cost rising too quickly, you can interrupt the agent by sending a stop command or by using the Escape key functionality through the mobile interface.

As a session grows longer, the context window fills up with the history of conversation and the files the agent has read. This can make the agent slower or less accurate. You can use the /compact command to summarize the session so far and free up space in the context window. Doing this from your phone ensures that the agent remains efficient throughout the day. If a session is no longer useful, you can use /clear to start over in the same folder without losing your configuration.

Customizing the remote terminal environment

The terminal environment itself can be optimized for better visibility on small screens. Using the /terminal-setup command allows you to configure how the interface looks and feels. You can also use /config to adjust settings like the model being used or the specific permissions for individual tools. These changes are saved in your local settings files, so they persist across different sessions and devices.

If you find that the agent is not using a specific tool correctly, you can check the status using /doctor. This command checks your installation and helps identify problems with MCP servers or environment variables. You can run these diagnostic commands from your phone to troubleshoot your setup without needing to sit down at your computer. By maintaining a healthy terminal environment, you ensure that your remote monitoring remains reliable and productive.

Questions people ask

Can I use the phone app to start a brand new session?

The phone app mirrors the existing wall on your computer, so you should have the folders and sessions already open in the desktop app. Once a tile is active on your wall, you can use the phone to send the initial prompt or provide the first set of instructions to start the work.

What happens if my computer goes to sleep while I am away?

The desktop app must be running for the mobile mirror to function. You should adjust your computer's power settings to prevent it from sleeping while you are running long coding tasks. If the host app closes, the phone app will lose its connection until the host is restarted and the link is re-established.

How do I manage multiple MCP servers from the mobile interface?

You can use the /mcp slash command from your phone to view and manage your connected MCP servers. This allows you to check which servers are active and see their current configuration. If a server needs to be restarted or modified, you can send the appropriate instructions to the agent through the mobile prompt.

Read next

More on claude code basics·All posts