An image of the terminal process installing Hermes Agent of Nous Research

So after just a few days of experimentation I decided to move from OpenClaw to Hermes.

Trying to stay up to speed with the developments; OpenClaw seems to be what we in Germany call “the snow of yesterday”. And I can relate:

In OpenClaw, the division between Heartbeat and Cronjobs is not necessary, and annoying

On top, in OpenClaw, there was just way too much stuff to do. Dreaming? I turned on Dreaming - and it seemed just a whole feature to feed into the hyped up narrative of the “sentient bot”. You easily get lost in these features.

Things break, the agent misunderstands it; the configuration is cluttered…

I am curious to what this will be. I am not expecting too much of the “self-learning”-thing with Hermes, though.

Things to take care of / Pitfalls

So, I saw that there is a command to migrate: hermes claw migrate

Knowing this, I went into the install. But the installer already has some migration capabilities:

Hermes install process prompting: OpenClaw installation detected

If you choose yes here, you will get conflicts in the migration process later on. So I advise against it.

So, let’s go:

# Full migration including API keys, skip confirmation
hermes claw migrate --preset full --yes

Hermes migration finished

Well… now we only still need to bring all the agents to hermes. I seem to notice: The Hermes mental model is more focused on one single orchestrator agent; not so much on parallel “main” agents? Hm, that’s kind of a bummer. I still need some time to confirm that. For now I will leave it be and return later.