Can Hermes Agent Use Claude Code? (Yes — Here's How)

Added: 2026-05-31

[00:00:00]Recently, I showed how my Hermes agent uses open code to write code. In the comments under that video, a question came up whether a Hermes could use Claude code in the same way instead of open code. I had never really thought about that before since I personally like using Claude code directly.

[00:00:18]However, I became curious whether such a setup was actually possible. So, I decided to look into it. And as it turns out, Hermes can indeed use Claude for writing code while Hermes itself acts as kind of project manager or high-level orchestrator. Setting this up is actually quite simple because the Hermes developers recently added Claude code integration directly into the Hermes agent skills system. So, it no longer requires any separate configurations.

[00:00:49]You might wonder how effective this combination actually is and whether it would simply be better to use Claude code directly. And in my opinion, if you need to work with code yourself directly, then using Claude code on its own is probably still the better option.

[00:01:04]However, if you want everything to happen with minimal involvement from your side, then Hermes becomes very useful. In addition, Hermes has long-term memory, meaning it remembers your previous projects, architecture preferences, and conversational context between sessions that can also come in handy. So, there is no universal answer to whether you should use Claude code together with Hermes. It all depends on the project and your specific tasks.

[00:01:31]Now, a few words about how to actually run them together. It's very simple. All you need is a Hermes agent as well as Claude code installed and configured.

[00:01:40]After that, launch Hermes and check whether everything is ready for this type of interaction. The easiest way is simply to ask Hermes whether it can use Claude code. It will check everything on its own and let you know. If everything is working correctly, then whenever you need to work with code, you can simply add something like use Claude code or do this through Claude code at the end of your request. After that, Hermes will automatically invoke Claude code with the necessary parameters. And once the task is complete, it will report the results back to you. By the way, Hermes [clears throat] can operate in two modes, interactive mode and print mode.

[00:02:18]Depending of the complexity of the task, it will automatically choose the appropriate mode. So, your involvement in is not required here, either.

[00:02:27]Overall, I find the workflow where Hermes controls the process and returns final report to be quite useful for some of my tasks. I will definitely continue using it from time to time in the future. However, as I mentioned earlier, this is not universal workflow, so use it carefully. And one more important point, in this case, you aren't using the Claude subscription anywhere else.

[00:02:49]It is being used specifically through Claude code. The only difference that it's Claude code is being launched and controlled by Hermes, rather than directly by you. So, in theory, this should not violate Anthropic's rules.

[00:03:01]However, if you decide to use this workflow for regularly, I would still recommend checking the Claude code terms of service to make sure there is no restrictions against this type of usage.

[00:03:12]That's probably it for today. Don't forget to leave a like and subscribe so you don't miss the future episodes. See you soon. Take care.