Last updated 13 October 2026
Monitor multiple Claude sessions on one wall
Managing multiple tasks creates cognitive load
Developers often jump between three or four different features or bug fixes in a single morning. When you use a command line agent to handle these tasks, you quickly run into a visibility problem. You start one process in one terminal tab and another in a different window. Soon you have five tabs open and no idea which agent is waiting for a permission and which one finished its task ten minutes ago.
Switching between these tabs takes focus away from the code you are currently writing. You have to remember which folder corresponds to which task and check each one manually. This constant context switching is the primary cause of fatigue when working with autonomous agents. You want the agent to work for you, but you end up spending your energy managing the agent itself and its individual states.
Claude Code sessions maintain distinct project contexts
When you run the claude command in a repository root, it looks for instructions in a CLAUDE.md file. This file contains the project specific rules that guide how the agent should behave during that session. Each session you start keeps its own history of tools used, like Bash or Read, and the conversation so far. If you need to stop a session and come back later, you use claude --continue to pick up exactly where you left off in that directory.
Managing many of these sessions requires careful organization of your local file system. If you want to work on a sub-module, you might place a CLAUDE.md in that specific subdirectory to give the agent more focused context. You can also use claude --resume to see a list of previous sessions if you accidentally closed a window. However, looking at a list of session IDs is not the same as seeing the work happen in real time across your whole project.
Terminal multiplexers often hide the current status
Many developers try to solve the visibility problem by using tools like tmux or screen. These allow you to run multiple terminal sessions in one window, but they still require you to switch between views to see what is happening. You might have a session running a complex Grep or WebSearch tool, but you cannot see the progress unless that specific pane is active. This lack of ambient information makes it hard to know when an agent is stuck or finished.
Even with multiple panes, the standard terminal does not give you a high level summary of what each agent is doing. You see the raw output of the command line interface, which is full of text and tool calls. If an agent is waiting for you to confirm a Write or Edit tool action, you might not notice for several minutes. If you encounter issues with a session, you can run claude doctor to check the installation status, but this does not help with monitoring the actual progress of your code.
The adhdTerminal wall provides a unified overview
The adhdTerminal app addresses this by turning each Claude Code session into a tile on a visual wall. This native app for macOS and Windows runs many sessions side by side. Each tile has a specific color, a number, and a name that you assign. You can see at a glance whether a session is working, waiting for your input, done, or idle. This means you do not have to click through tabs to find the one that needs attention.
The wall shows the last few exchanges of each conversation directly on the tile. You can see the headlines or numbers the project publishes without opening the full terminal view. If you restart your computer, the tiles reappear in the same place and resume the sessions. This persistence is helpful for long running tasks that span multiple days. It provides a stable environment for your agents to work while you focus on the most important parts of the code.
Project instructions shape session behavior
To get the most out of a tiled setup, you should use the CLAUDE.md file in your repo root or subdirectories. This file tells the agent how to build the project, how to run tests, and what style guides to follow. When you have multiple tiles running, having clear project instructions ensures that each agent follows the same rules. You can also use a global instructions file at ~/.claude/CLAUDE.md for rules that should apply to every session on your wall.
Settings are also important for managing how the agents interact with your system. You can define permission modes in .claude/settings.json or .claude/settings.local.json. The settings.json file allows you to define per tool permissions. You can specify which tools like Bash or Edit require a manual confirmation and which ones can run freely. This granular control is vital when you are managing a wall of multiple agents, as it prevents accidental system changes across several directories.
Mobile monitoring keeps long tasks visible
When an agent is performing a large task involving many Glob and Read operations, you might want to step away from your desk. The adhdTerminal system includes a free iPhone app that mirrors your wall over an end to end encrypted link. This allows you to check the status of your tasks from another room. You can see the same tile colors and status indicators that you see on your desktop.
The mobile app is not just for viewing. You can answer a prompt or send the next task to an agent directly from your phone. If an agent finishes a task and is waiting for your feedback, you can provide it immediately without going back to your computer. This keeps the momentum of the project moving forward even when you are not actively typing at your desk. It reduces the overhead of managing long running background tasks.
Controlling context and costs across sessions
Running multiple sessions simultaneously can quickly fill up the context window. Each session tracks the files it has read and the conversations you have had. When the context window gets too full, you can use the /compact command. This summarizes the session so far and frees up space for new information. This is particularly useful for tiles that have been active for a long time on your wall.
You also need to keep an eye on the resources you are using. You can use /cost and /usage inside any session to see how much you have spent. Since a visual wall shows many sessions at once, it is easy to forget that each one incurs its own costs. Monitoring these numbers regularly helps you decide when to use /clear to start a session over or when to finish a task and close a tile to save resources.
Permission modes ensure safety during parallel work
Safety is a major concern when multiple agents have the power to change your code. Claude Code uses several permission modes to control what the agent can do. The default mode asks for permission before using tools like Write or Bash. If you are monitoring many tiles, you might use the acceptEdits mode for tasks you trust, which allows the agent to edit files without asking every time. You can switch between these modes easily using Shift+Tab within each tile.
For more complex planning, the plan mode allows the agent to think through a problem without actually executing the changes. If you need to move quickly, you can use the bypassPermissions mode, but this should be used with caution. It allows the agent to execute any tool without your intervention. If an agent starts doing something unexpected, you can hit the Escape key to interrupt it. Pressing Escape twice allows you to go back to earlier messages and fix the prompt.
Questions people ask
How do I change the name of a tile on the wall?
You can set the name of a tile within the adhdTerminal interface to reflect the specific task or branch you are working on. This name stays persistent even after you restart the application or your computer. It helps you distinguish between different repository sessions at a single glance.
Can I use different MCP servers for different tiles?
Yes, you can configure MCP servers in the .mcp.json file within your project directory or use the claude mcp command to manage them. Each session will respect the configuration found in its specific environment. This allows you to have different tools available for different project tiles on your wall.
What happens if I lose my internet connection while sessions are running?
Claude Code requires a connection to reach the model for processing. If the connection drops, the agent will stop its current task and you will likely see an error or a paused state on the tile. You can resume the work once your connection is restored by using the resume functions provided in the interface.
Read next
- Add build instructions to your CLAUDE.md fileImprove project context by adding specific build commands and environment requirements to your CLAUDE.md file for the AI to follow.
- Add common error fixes to CLAUDE.mdImprove project memory by adding a section for common errors and their fixes to your CLAUDE.md configuration file.
- Add custom shell aliases for Claude CodeSpeed up your workflow by creating shell aliases for common Claude Code invocations and project-specific startup commands.
- Add test commands to a CLAUDE.md fileUnderstand how to define specific test runners and environment flags in your CLAUDE.md file so the tool runs the correct test suite.