coding agents sequentially and not with multiple runs in parallel, you lose. One of the key advantages of coding agents is that you can start completing tasks in parallel, something that was never possible before when working on software engineering tasks.
However, when I start using many of the same coding sessions, it often becomes difficult to keep track. You need special techniques to easily keep an overview of all the sessions you run, quickly grab conversations when you return to them, and more.
Why is it difficult to use coding agents in parallel?
First, I want to cover why it is challenging to use coding agents in parallel. To some extent, this should be self-explanatory. Before LLMs, software engineers and programmers would only work on one job at a time. The simple reason was that if you try to multitask in software engineering jobs, you will end up being inefficient. The reason is that software engineering jobs are often complex and require your full attention. If you try to do other things at the same time, performance in all tasks may be interrupted.
However, that game has changed a bit, as you no longer write all the code yourself. At least in my opinion, you shouldn't write all the code yourself anymore, because coding agents should write the code for you. As a programmer, you must now act as a manager of coding agents, instead of writing code yourself.
However, if you are a manager of coding agents, you naturally have to manage coding agents that perform different tasks. Running tasks in parallel naturally requires tasks to be unlinked. This will again present the same challenge as before, where you are working on programming tasks in parallel, and you need to keep a lot of context in your working memory, and you need to know, for example.
- Answer your coding agent whenever they ask you a question.
- Check the code agent's usage after it has finished. You need to know what to check, how to check, and how to make sure the writing agent is doing what it's supposed to.
How to effectively use multiple agents to install the same code
In this section, I will cover some techniques that I use and use every day to successfully use multiple coding agents.
Agents look at Claude Code
An effective way to get a more comprehensive view of your agents is to use the agent view. For example, the agent view found in Claude Code. Many different providers have different ways of presenting this. I know that Warp, the terminal, also recently introduced a new way to view multiple different agents.
The good thing about the agent view is that, as you can see in the image above, you don't need to fully display every conversation you have with the agent. It's just one line, which you can press enter on if you want more details on it. Otherwise, it will just be a background task, and it will ask you to input anything when Claude sessions ask you for input.
I think this is a very efficient way of working, as you can easily have multiple coding sessions running at the same time without confusing which agent is running, which agent needs input, and so on. You can activate the agents view in Claude Code by:
claude agentsWarning if the code agent requires input
Another important thing you can do is to be alerted whenever coding agents need input. Of course, if you're using agent mode, you can have this, as it's clearly marked for you whenever the code agent needs input from you. However, if you don't like the agent view or want to use something else, there are different options.
In the image above, you can see that I have different instances of Claude Code running in different tabs of the terminal. The tab includes an asterisk between the tab title and the Claude code icon if it requires input. This is the easiest way for me to see which terminal tabs I need to do something with and which tabs are running in the background.
Another thing you can do is also have an audio signal whenever the agent needs input from you. You can, for example, do this by using hooks in Claude Code, which are processes that run at specific points in time. A hook, for example, can be triggered every time Claude needs input from you, and you can connect this hook to an audio signal that plays, so it notifies you whenever one of your coding agents needs input.
Enable shortcuts
Recaps are another incredibly powerful feature that you can use to effectively use multiple encoding agents. A common problem when using multiple agents is that it is difficult to find the context of a particular agent.
Also, suppose you have five agents working in parallel. You first deal with agent 1 and tell it what to do, and so on, then spin agents 2, 3, 4, and 5. At that point, it's probably been 10 minutes since you interacted with the first agent, or more, and you need to re-contextualize it again: what were you doing with that first agent, and what were you trying to accomplish, and so on. This can be difficult if you don't have a recap or similar, but this is where the recap feature in Claude Code comes in very handy.
The recap element, as you can see in the image above, is simply a piece of text above the user input field. It summarizes what you have been doing in this thread and what you are trying to achieve. You can easily read that text, and you will quickly respond to the context and be able to communicate with your agent.
New tabs or split panes
The last method I want to cover in this article is new tabs or split panes. I highly recommend that you work in terminal view or another platform that allows you to split panes when working with coding agents.
The image above shows an example of a split pane. If you're working in a Warp terminal, you can press Command+D on your Mac, and it will split the current view horizontally so you have two terminals you're working on. This is very powerful because it allows you to quickly have an overview of two agents at once, which I find very useful.
There are probably many terminal providers and other coding agents that allow you to have this split-pane setup, but I highly recommend that you find one that works for you.
Also, the way I like to work is that I have a new tab when I'm working on different folders, so I have one tab per folder. If I work on multiple agents in that folder, I split the panes.
This allows me to quickly get an overview of my coding agents running on various repositories.
The conclusion
In this article, I have discussed how you can keep an overview of the same coding sessions. I discuss the different techniques I use to successfully run multiple agents in parallel and continue to look at their pros and cons. I believe that the future of programmers is that they will be orchestrators of code agents. You will be the manager of an AI agent. You should immediately start working on the intelligence of working with coding agents in conjunction with managing many of them, as I believe this will be an incredibly important skill in the future if you work as a programmer.
👋 Touch
👉 My Free eBook and Webinar:
🚀 10x Your Engineering with LLMs (Free 3-Day Email Course)
📚 Get my free ebook Vision Language Models
💻 My webinar on Vision Language Models
👉 Find me on social media:
💌 Stack
🐦 X / Twitter
