YouTube

Hermes Agent is insane… 100,000+ github stars

32:47English129 segments6,578 words · 33 min read

Search inside any video

SavedThat transcribes your saved videos and lets you search across all of them instantly. Save this video and find any moment.

TL;DR

In this video, David Ondrej explains Hermes Agent, its rapid growth, capabilities, and how to set it up for various tasks, including browser automation.

Hermes Agent featureshow to set up Hermes AgentAI agent capabilitiesHermes Agent use casesbrowser automation with AIGitHub stars Hermes AgentAI in marketing and educationself-improving AI tools

Chapters

  1. 0:00Introduction to Hermes Agent
    01
  2. 1:00Growth and Popularity
    02
  3. 2:00Use Cases of Hermes Agent
    03
  4. 5:00Setting Up Hermes Agent
    04
  5. 15:00Using Browser Harness with Hermes
    05
  6. 25:00Final Thoughts and Conclusion
    06

Transcript

0:00

My name is David Andre and this video is the clearest explanation on the internet on what Hermes agent is, what it does and how to use it. First off, you have to realize that this is becoming the fastest growing and most popular AI agent in the world. Just look at this trajectory of GitHub stars. And in fact,

0:15

trajectory of GitHub stars. And in fact, it might be the end of OpenClaw. And that's not just my opinion. Let's look at the data. This is Google Trends, aka the search history on these two projects for the term Hermes agent worldwide past three months. we can see that it's growing faster and faster. But for

0:30

growing faster and faster. But for OpenClaw worldwide past three months, we can see that the growth has peaked in late March and now it's steadily declining. So this really might be the open claw killer. Now what people don't realize is that Hermes agent is the fastest project in all of GitHub history

0:47

to reach 100,000 stars. And that's mainly because of the insane update speed. competing with this team is insane. five major releases in the last 20 days alone with over 741 merge PRs. That's around 37 pull requests per day. Now, you might be

1:03

requests per day. Now, you might be thinking, "But David, what can Hermes mess do? , what is it used for?" Well, let me show you. The first use case is insane. So, this is Ply. She used Hermes to jailbreak Gemma 4. The crazy part is it was done with

1:16

4. The crazy part is it was done with Hermes agent and a single skill called Obliterus with only eight prompts from a human. , Hermes agent figured out how to jailbreak this model by itself so that it can answer any question imaginable. But let's look at a more practical use case. This is what

1:31

more practical use case. This is what Adam did. He used Hermes agent to build a video in Mandarin. So the language that they used in China where it wrote full HTML file, generated Chinese text to speech, rendered a vertical 916 video 1080p and gave the full MP4 file. Look

1:48

at this. First of all, this is going to be huge for both marketing and education. this really might be the best way to learn a new language. Another crazy use case for Hermes agent was building this competition entry for the actual Hermes

2:03

competition entry for the actual Hermes agent creative hackathon. Here it created graphics of real sculptures and instead of slop which you would expect from AI. It created these animations, these GIFs which have a certain vibe,

2:20

certain feel. This is not the typical AI slope feel. This feels a lot more custom, a lot more high value, a lot more branded. So this is the type of stuff that Hermes agent can do. But before we set up Hermes agent, which of course I'm going to show you later in the video, here is how to make it near AGI level. And what by that is

2:35

AGI level. And what by that is that it can complete any task on the computer that a human can. And that is thanks to the new browser harness GitHub repo. Now, this is a GitHub repo that many people have missed because it's very new, just couple days old. As you can see, it's under 2,000 GitHub stars, which I don't expect it to

2:51

GitHub stars, which I don't expect it to stay so underground for so long. But this is the PI agent version for browser use. As it says, it's the simplest, thinnest, self-healing harness that gives AI models complete freedom to complete any

3:04

models complete freedom to complete any browser task. In other words, if the agent cannot do it with the default library, it can create new functions and write new skills to complete any type of task on the internet. In fact, the creators of browser harness, which is

3:18

creators of browser harness, which is the team behind browser use, are so confident in this that they said the first person to find a task that doesn't work with browser harness will get a fresh Mac Mini. So, if you're watching this, you can get a fresh Mac Mini if you figure out where browser harness doesn't work. So, of course, in this

3:32

you figure out where browser harness doesn't work. So, of course, in this video, I'm going to show you how to set up Hermes agent with this new browser harness tool. If you don't really understand what browser use is, think of it as a tool that lets the agent see, click, and type on any website just a human can. Simply put, Hermes agent is

3:50

the brain. Browser use is the hand. So, here's example how it works. You can tell it to go to a specific website and do some actions on there such as go to Twitter and post a tweet or go to Amazon and buy something or go to hacker news and scrape the latest hacker news or go to LinkedIn and get you some leads.

4:06

to LinkedIn and get you some leads. Whatever actions you yourself are doing on the computer, now you can delegate them to an AI agent and thanks to the browser harness tool, it can do them. Okay, so now let me show you how to set this up. First off, we need the VPS because we want our agent

4:22

running 24/7. And the best and most affordable option I would say is Hostinger. They have a pre-built page for Hermes agent. just go to this page which I'm going to link below the video and click on deploy and set up a new VPS that is already comes

4:36

set up a new VPS that is already comes with Hermes agent. So you don't have to do any of the advanced setup. They made it so much easier to just start running. So first off, select your period. I recommend going either 12 months or ideally 24 months because this is the best deal possible and you want your AI

4:51

best deal possible and you want your AI engine running forever. That's the beauty. You sit down for 60 minutes for 90 minutes. You set it up once and it's saving you time for the rest of your life. Now, if you want an even better deal, go to the and click on coupon code and type in code David for additional 10% off. Yeah, this is a

5:06

additional 10% off. Yeah, this is a pretty good deal. Shout out Hostinger for sponsoring this video. Now, scroll down and select the location here. Make sure it's the closest to you. So, I'm Czech Republic, so Germany is super close. And then just click on continue. Now, the last thing that remains is filling out your billing address and your payment details. And then click on

5:20

filling out your billing address and your payment details. And then click on submit payment. So, again, if you want to set up Hermes agent yourself, make sure to do it on Hostinger VPS. The link to this page is going to be below the video. All . Then you'll get redirected to the hostinger panel where you can manage all your VPS servers. All . As you can see, the newest VPS is

5:35

. As you can see, the newest VPS is this one. I just bought it. KVM 2 plan 24 months. Boom. You can see that I have two more years. And then let's click on setup. Confirm the location. Click on next. Now here, don't select an OS. Instead, select an application. Make sure it's the Hermes agent. Select. So, , select the username and save

5:51

, select the username and save the password. , just make sure to save this and then click deploy. And there you go. your VPS is starting to deploy. Usually takes two minutes. , they say five, but usually it's one or two minutes. And then you're going to have your own virtual private server, which is far more than most people will ever reach.

6:07

more than most people will ever reach. And the beauty of having your own VPS is that you can run multiple agents on there. So if you want Hermes agent, if you want cloud code, Codeex, OpenClaw, agent zero, if you want to run multiple agents on a single VPS, you absolutely can. The KVM2 plan is more than strong

6:21

can. The KVM2 plan is more than strong enough for this. Okay, there we go. Everything is set up. took two minutes, maybe three minutes. And we can see that we have multiple Docker containers, two of them. This is the default one, but we care about this one, the Hermes agent one. So, click on open. And here we need to log in with

6:37

open. And here we need to log in with the same credentials. So, obviously the Hermes was the name, but hopefully you saved the password. So, paste that in here. Boom. And click on sign in. And there we go. We are inside of the Hermes agent setup. Obviously here we can set up which model we want to use and most importantly what we want to use for the

6:53

importantly what we want to use for the browser use because that's where we going to set browser harness. So let's go with a quick setup. You don't have to go with the full setup. I'm going with open router. This will allow us to use any model we want. So let's click on that. We need to add our open router API key. So just go to open router.ai. And

7:08

key. So just go to open router.ai. And once you're at open router.ai just click on get API key. If you don't have an account by the way create one. It's completely free. And open router is one of the most essential things to have in AI. Whether you're building software or whether you want to run your personal

7:23

whether you want to run your personal agents. So just create an open router account. Trust me. Now go to the left and charge up a couple of bucks. You don't need 77. Just do5 or $10. That's more than enough. Go to the API keys. Click on create. I'm going to name it YouTube Hermes VPS. I'm going to put

7:41

some limit on it $30. This is a good practice. So in case it's leaked, I don't have to worry about it. Now, obviously I'm going to revoke my API key after this video. Do not share your API keys with anybody, okay? Treat them as passwords. So, copy this. Boom. Jump back into Hermes setup. Paste it in. Now, it will not be visible. So, don't

7:57

Now, it will not be visible. So, don't be confused by that because when you paste something into terminal, it's a secret. It doesn't show you the characters, ? So, you just make sure to click and then do Ctrl +V to paste. Okay. So, here we need to select the default model which should be as strong as possible. Obviously, there is

8:11

open 4.6 4.6. But there's a newer model which is Opus 4.7. So if we go to open router and type in Opus, you can see that we have two options. If you have plenty of money and you want 4.6 fast, which is six times more expensive, then go with that. Or if you want the latest

8:26

go with that. Or if you want the latest model at the normal pricing, just go with Opus 4.7. Just make sure to copy the name of the model, ? And again, inside of open router, you can use any model. They have open source models, openi models, an entropic models, google

8:41

openi models, an entropic models, google models. All of the models that exist are on open router. That's why I use it. Okay, so let's go back and here press arrow up to go to the bottom and press arrow up again to be on the enter custom model name. Enter that and paste

8:58

in. Okay, I copied that. My bad. So let's go back and we need to paste in the opus 4.7. So copy this model name. There's a copy button here. Let me delete this and paste in this. There we go. Now, do we want to set up messaging? Now, honestly, you

9:13

set up messaging? Now, honestly, you probably should if you want to use it through Discord, Telegram, WhatsApp, Slack, but I'm going to skip it for now because the focus on this video is on the browser harness. Boom. Do we want to launch Hermes chat now? , no. We don't want to launch Hermes chat because

9:26

launch Hermes chat now? , no. We don't want to launch Hermes chat because first we want to install the browser harness skill. So, I'm going to type in N for no and hit enter. Now here we are inside of Hermes. It still started it and we can start chatting with it. As you can see I type in hey and Hermes

9:40

you can see I type in hey and Hermes agent is responding with cloth open 7. So this is already a big win ? This allows us to already use Hermes agent and its self-arning abilities. That's why it's popular. This is something that openclaw doesn't have the ability to create skills by itself. It's

Keep reading — 90 more segments

Sign in free to read the full transcript, save this video, and search inside everything you save.

Sign in to continue reading

Prefer the original? Watch the video

Are you the creator or rights holder of this video? Request removal of this transcript.

Related Transcripts

Never lose a moment again

Save videos from YouTube, Instagram, and TikTok. Search across all your transcripts with AI-powered semantic search.

Start Saving Videos — Free Trial