4 FREE Ways to Save 80% on Claude Claude Tokens!

Today, I'm going to show you four free GitHub repos that cut Claude code tokens by 80%. So this is a full token minimization playbook that cuts your Claude code tokens by 80% and all of it is free. Now you can use this with any AI agent. It doesn't have to be Claude code.

And these are four free tools, each one reducing the amount of tokens you use so you can get more out of your AI agents. That means your AI agents get faster. Your limits stop running out and everything keeps working with the setup you already have. And I've actually tested every single one.

In fact, one of them actually cut a single command's output by 92%. We'll come on to that later. So that by the end of this, you'll have the whole stack installed and your tokens will go five times faster, five times further. So the problem with Claude right now is like most of the time we're using tokens without being smart, right?

So Claude gets flooded, you get like giant test logs it didn't need to read. You've got polite three paragraph answers, overbuilt code nobody asked for, right? And also like four jobs being done, which a free model could actually do for you, right? So it takes four different tools and all three are free, open source repos that you can use to get more out of this stuff, right?

And so if you're wondering, okay, what are the problems with this? Well, leak number one is tool output. And so one of the biggest problems is the Claude replies way too politely with loads of filler words, which reduce, which increases your output tokens, which uses up your token plan. Also, another problem is overbuilt code, right?

So sometimes it overbuilds the code that it gives you and it's full of fluff that it doesn't need to. Additionally, you can end up paying like full price for grunt work and it renames variables and write stuff that it doesn't need. And also tool output, right? It sometimes floods with context.

So these are all things, all different leaks that sneak into Claude and cost you more tokens. And so you might say, well, is this actually worth the whole setup time? The whole stack can install in about 15 minutes with one command each. And after that, it's pretty much automatic.

You don't need to do it again. We've actually got it running inside our agent operating system. So for example, with Claude, we can use these token minimization playbooks to reduce the amount of tokens we use each time. So what is the token minimization playbook?

Well, there's four parts to it. So we have RTK, Caveman, Ponytail and Omniru. Now, RTK is basically a powerful way to reduce the tool output and it can actually reduce 89.9% based on my own tests. It reduced the token usage by 82.9%.

Basically, what this means is like when it is running shell commands, usually it needs a fraction of what it actually uses. And the problem is like if it's using a git diff, which you don't need to know about, but basically it's a way of doing shell commands. If it's messy, it can use up thousands of tokens. And so RTK sits between the agent and the shell command, which rewrites the command to compact it, filter the output, deduplicate the noise and hand Claude only the signal.

It adds about 14 milliseconds, so it doesn't slow it down. You don't see it working. It just reduces the amount of tokens you use. Now we actually tested it because on RTK, they claim 90% fewer tokens.

And actually on my tests, it reduced 82.9%. It was very, very accurate and very close. It's a free open source project. This is RTK, as you can see right here.

Basically, it allows you to reduce the amount of tokens we use, right? Now we've already got it built into the agent OS system with all of our agents running together. So we have Claude ready to go to reduce the amount of tokens we use. You might also say, well, won't filtering the output break the agent?

So it only filters mechanical noise, right? And that's all it's filtering out there. Then we have Caveman, right? Caveman is the second part of the system, the second free repo.

And Caveman basically reduces the output tokens that it speaks with, right? So it can actually reduce tokens by 65%. So you can see an example right here. And if you look at the before and after, a normal agent would use about 69 tokens on an output.

Whereas for example, with Caveman, we reduce that to 19 tokens, which means that we use way less tokens per output. And instead of these fluffy filler responses that are super polite, it just gets straight to the point, which is what we want. And we use less tokens that way as well. And from my tests on Fable 5, it used 69% fewer output tokens, which reduces the overall tokens I was using by 30%.

And all of those answers were still technically correct. So it's not reducing the quality of the answers. It's just reducing the fluff from those answers. And we actually use it inside the agent OS too.

Now, the free repo number three to reduce your token output is something called Ponytail. And Ponytail basically reduces the amount of code that you write. So it's way more efficient. It basically acts like a lazy developer, which means you use about 54% less code.

It's 20% cheaper, 27% faster, and really, really powerful, right? So really helpful for reducing the amount of tokens you use with Claude. And you also might say, well, I don't want AI cutting corners on my code. But Ponytail isn't touching, for example, like safety stuff or anything like that.

Not even, for example, like input validation or error handling. It's just reducing the speculative stuff, the abstractions for later that later never needs. And so it's simplifying the code that you create. So you reduce the amount of tokens you use so that you get more out of your Claude subscription by creating less code, about half the amount of code.

And that's all like the unnecessary code. So you want to see Ponytail as like a lazy senior developer who just creates less code, but still gets the job done. And then finally, we have Omniroot. Now, Omniroot is a free coder.

And what you can actually do is use Omniroot to delegate like the grunt work, to delegate the tasks that don't require a frontier model like Claude. Now, Claude can actually use this directly. Or we actually have a free AI coder that runs with NineRoot and Omnirooter, so that for those little tasks that don't require a frontier model, you can plug that in. And so this is a local gateway that runs on your machine with the agent OS, it has 93 models.

And it actually has RTK and caveman compression built into the gateway itself, which squeezes another 15 to 95% out of every request that passes through. And so this is a really powerful system for using free models and avoiding getting rate limited on those free models, because you've got RTK and caveman built in. And so this cuts a lot of the total output, right? RTK alone reduces 82.9% of the total output.

Ponytail reduces about 50 to 20% of the code volume. You've got, for example, caveman reducing the reply tokens by 69%. And then you've got Omniroot, which completely reduces the amount of tokens you use because it's a free API. And you might also say, OK, what's the catch?

Well, I mean, it's not going to work on 80% of everything you do. But if you do want a more efficient way of using it, then this is better. And also, like, bear in mind, when you add them all together inside one system, it might be more than 80% of the tokens you use. And so before, I was burning through, like, Claude code limits all the time.

You know, every session would start fast, then it would stop. You know, the context would get bloated. The output tokens would get bloated. But with this system, this stack that runs my whole agent OS, RTK filters every shell command.

Caveman keeps every reply tight. Ponytail keeps the code base lane. And also Omniroot sends the grunt work to free models. And so we get the same quality of answers, but I can run more agents.

I can build more. And I don't have to worry about, I don't have that token anxiety, essentially. And you can also do a bunch of other stuff, right? I can give you eight more tips.

So another thing that you can do is you can run forward slash clear if you finish the task. And this means you start with fresh with old context removed, which reduces you out of bloat. So you can use clear on a finished task. And also if you're on a long single task, you can run forward slash compact to summarize a history instead of dragging it all forward.

Because the big problem with Claude is like when you're using it in long conversations, it bloats up with all the context from before. You can also put Claude MD on a diet. So Claude.md is the instruction file that rides along with every session, every state, every rule. You can delete like half of it, right?

And nothing will break. Like Claude actually put out a new guide recently where they actually said this system, the whole Claude code system, they reduced the code by like 80% and still got the same quality of outputs. We've also got a memory vault inside our agent OS, and that just allows us to stop having to re-explain the context every time because it reads from our memory and it understands everything about us. So that reduces a lot of context as well.

So there's all these different ways to reduce these systems. You can also route by difficulty. So you can give like the simple task to cheaper models like local or Omni route and hard stuff to the frontier model. If you plan before you build, you also reduce tokens.

If you batch your asks instead of doing the one by one, that can reduce the tokens as well because you're not using the same context three times over. And also you can also have a sub agent look through a code base for you. Right. So you can use a scout and the scout basically burns its own context and reports back.

So your main session only sees a conclusion, which saves a lot of token usage as well. So the old way is that you treat tokens like a scarce resource, you know, shorter prompts, fewer questions, anxiety before every task, you hit the limits quickly. The new way with this system is that you fix the plumbing once you've run these four free repos to filter the noise, shrink the replies, thin the code and route the groundwork to free models automatically. And then you can get more out of your AI, not less.

Right. And that's the whole point. So if you want this whole system, we've actually plugged it into the agent OS system already. So you can get it inside our agent OS.

It also comes with all of our agents working side by side. As you can see right here, we've got tons of cool custom workflows like a video agent, SEO agent. We've got, for example, Gemini notebook plugged into the system, obsidian with a memory galaxy, as you can see right here. And you can, this means you can run an agent operating system in a way more efficient way, right?

So your agents run smoother and faster. Now, if you want to get that full system is inside the AI profit boardroom link in the comments, description, or go to the air profitable.com. So you can grab the agent OS inside the classroom. Just go to this section over here.

Um, we update this daily with a new, we've got a video tutorial right there. You can see the last update day and you get zip file to install it. And then we add new tutorials and guides all the time as well to help you as much as we can. You can post inside the community and get your questions answered by me personally in a video tutorial.

And then inside the calendar, you can jump on weekly coaching course. If you're also worried about tokens, then you can check out the section on how to minimize tokens. So we have a full token minimization playbook here. And then also you can learn how to run an agent operating system for free with this system right here.

All right. So everything you need to win with AI is inside the air profit boardroom. And then also you might be wondering, okay, is it technical to use this sort of stuff? Absolutely not.

Right. So we've got 205 pages of wins of testimonials from people inside the AI profit boardroom community and people absolutely loving it. Right. So I'm non-technical then on technical, but you can still get amazing results for this stuff.

If you have the right systems, which is what the agent operating system is all about. So thanks for watching. Hope to see inside that link in the comments description, or just go to the AI profit boardroom.com to get access. Thanks for watching.