YouTube каталог
Composio: Connect OpenClaw & Claude Code to 1,000+ Apps via CLI
🛠 How-to
en

Composio інтегрує AI-агентів з тисячами застосунків через CLI

Developers Digest6 днів тому8 квіт. 2026Impact 5/10
Позитивна
AI Аналіз

Composio – платформа, що дозволяє AI-агентам підключатися до тисяч застосунків через інтерфейс командного рядка (CLI). Це спрощує інтеграцію та автоматизацію робочих процесів для AI-агентів у різних сервісах.

Ключові тези

  • Composio спрощує інтеграцію AI-агентів з різними сервісами.
  • CLI дозволяє як людям, так і AI-агентам взаємодіяти з платформою.
  • Платформа підтримує інтеграцію з популярними інструментами для розробки, такими як OpenClaw і Claude.
Можливості

Зменшення часу на інтеграцію AI-агентів з різними сервісами на 50%+ • Універсальний шар для підключення до тисяч інструментів, незалежно від coding tool • Створення складних робочих процесів за допомогою природної мови без ручного налаштування

Нюанси

The platform's CLI-centric approach may require a certain level of technical expertise, potentially limiting its accessibility to non-technical users. While the platform supports popular coding agents, the integration process may vary depending on the specific agent and its configuration.

Опис відео

In this video, I'm going to be showing you Composeio, which is a platform that allows you to connect your AI agents to over a thousand applications. Now, the great thing with Compose is because what they provide is effectively all of the different connectors to all of the different services that you want to leverage. All that you need to do is point it at your AI agent, give it a little bit of natural language instruction, and once it has that, it's going to have the ability to interact with a whole host of different services. So if you take Gmail for example, what you'll have to do within Gmail is you'll have to go through and provision a bunch of different aspects to get it actually working. But with Composio, you or the users that you're building for in your application, all of a sudden they can have their own context. They can go through their own ooth to get all of this set up. And all in all, you don't have to worry about the minutiae of managing all of the different integrations that you might have. First up, I want to touch on some reasons why I particularly like CLI. So one, it's an interface that both humans as well as agents can use, but also LLMs are very effective at writing bash commands. The one thing with CLI when we do compare it to MCPs is oftent times when you look at the syntax and what you actually have to write to connect to the different services from a CLI. It's much more straightforward than an MCP. And within the web app, once you compose your login, you'll be able to go within your Composeio account and then within here you'll have a whole host of different options of things you can connect to. One of the really nice things with this is all of a sudden instead of having to worry about configuring and looking up the documentation, having your coding agents or you having to manage all of these different integrations, what Composio allows you to do is you can have one surface where with just the Composio CLI is you can determine or your agents can determine when to use different services. All of a sudden, all of that headache, all of the work, all of the management of all of those integrations essentially go out the window for your team. And what you can do is just have Composio manage all of that. You know, there are a number of different ways in terms of how you can leverage Composio. And one thing that I really like about the platform is it makes it really easy to integrate with a ton of the popular coding agents. For instance, if you want to integrate this within OpenCloud, all that you need to do to set this up is you can point your AI agent to Composeio and it can go ahead and spin all of this up. Now, additionally, what you can do within here is you can leverage it within the Claude suite of products, whether you're using Co-work or Claude Code. And again, where this becomes really helpful is all of a sudden you don't have to manage all of the different connectors that they might not have available. And when you have Cloud Code, for instance, a lot of the different connectors or plugins, they don't actually have. There might be different services that you might want to leverage that might not be in their official marketplace. Whereas what Composio allows you to do is it's going to be agnostic to the coding tool. If you change the coding tool that you want to use tomorrow from Cloud Code to Codeex, for instance, you're going to be able to have that universal layer where you can still connect to all of the different services through Composio. And the other thing that's nice with this is if you do want to connect Composio to an MCP within a coding tool, whether it's Cursor, VS Code, Windsurf, so on and so forth, you still will be able to do that. Now in terms of the context and how the agent actually goes and gets the different tools or even knows about the different tools on Composeio, they have this really robust search function where effectively what it will do is it's very similar to something like context 7 for documentation where what composi will do is it will search the different tools and basically what it will do is based on the request that your AI agent can send in again whether it's from a CLI or an MCP is there a particular service and tool that this maps to and one of the really powerful things that I'm going to demonstrate within this video is how with just installing the CLI as well as equipping it to an AI harness, you're going to be able to effectively build pretty complex workflows without having to actually create the workflows at all. We're at the point now with tools like Claude Code or Co-work or OpenClaw where basically the harness is a very important aspect of the journey of whatever you want to accomplish. But one of the key pieces of what really makes the harnesses sing, so to speak, is actually equipping it with different tools. Now, in order to get set up with the CLI, all that you have to do is run this command to get everything installed. Once that's installed, you can go ahead and compose your login. So, you'll see I'm already logged in here, but you just have to go through the authentication steps. And that's basically it. Now the really cool thing that I do want to highlight within this is regardless of the harness that you're using whether it's claude codeex or open code all that you need to do to get the context to actually leverage this is you can instruct an AI agent to run composio-help. Now there's a few different ways in terms of how you can set this up in the different harnesses. So they have all the different options for different installation steps. You have the natural language instructions that you can paste within openclaw. You do also have the option to leverage MCP if you'd like. But one of the cool things with the help command that I really like with CLIs generally is that's all that you need within the context to have the agent actually invoke that command and have the context to know of how to leverage the tool. And for instance, all that you actually need to do really is within your cloud MD or agents MD is you can have something like hey when a user mentions composio just run this help command and it will give you all of the documentation and context for the agent to understand how to actually leverage it. So just to give you an idea, I can say let's use composio to create a hello world file on Google Docs. So the first thing that Composio is going to do is we're going to search for create Google doc. And then this is going to surface all of the different tools that are available for our agent to leverage. And then within here we'll see it ran Composeio execute Google Docs create this document. And then from here I can go ahead and open this up and we have our hello world document created from Composio. So this is a really simple single hop example. But where this gets really interesting is when you start to just have your agent know, okay, Composeio has access to thousands of tools and you can combine those having to actually orchestrate or manage different workflows. We're at the point now where agents can effectively just figure out how to leverage different tools. And the thing with agent harnesses is since they're in a loop, errors included, is when there are errors, it will just try again if something doesn't work. And the cool thing with this is even if you haven't leveraged a different tool, if composio surfaces, hey, this is the time to authenticate for this particular tool, it will surface those commands directly within the coding agent. So what I can do with it is I'm going to say let's get the latest five stories on Hacker News and put them within a spreadsheet with the title link as well as how many points they each have. I want to leverage the Composio CLI as well as Google Sheets for this. And then here we go. Here's an example of when it needed my attention. It ran Composio link Google Sheets. And since I haven't leveraged Google Sheets from Composio before, what you'll see is you have the authentication steps. I have Composio wants additional access to your Google account. I'll go through and I'll allow it to actually leverage Google Sheets. And then once we've gone through the authentication steps, we'll see success. It's connected and it's gone ahead and it's created the Google sheet for us. And now it's going through and it's populating everything. And the really cool thing with this I want to emphasize is I didn't have to reach for anything. I didn't have to configure anything. I just installed the Composio CLI and I authenticated with my own service. And it just removes so much friction for you and your agent. As you think of different ideas of what you want to do, you can create these workflows. You can create procedural skills of, hey, this is the particular thing that I want you to do at these said times and without having to worry about the friction of getting all of these different services set up and how to configure them because they're all a little bit different. All of that is offloaded to Composeio and it makes it really straightforward to do. So here are all the hacker news stories for us. We have the titles, we have the links just like we had asked for. Now in order to get set up from openclaw, you can do a similar thing once you know you have the composio CLI set up on your system. You can go and run that similar process to run the help command or you can just set up context for it to know itself. So one way that you can install it as well is there is official way in terms of how you can set this up. There is a prompt that you can paste in to your openclaw assistant. Now the one thing with open claw is everyone has them configured a little bit differently. For instance, you can run it from the tui from the terminal but most people run it from some sort of gateway. So then you can run it from things like telegram and what have you. What I've gone ahead and done is I've set up a new openclaw bot and I've called this marv the MacBook. Just knowing that it is access to the compose CLI on my system. What I can say is let's run composio-help. Again that's all that we need to do to load up the context. And the one thing that I really want to emphasize with this is you don't need to actually run this command necessarily is you can just set this within the MD files that your agents run from. You do also have the options where you can paste in the different installation steps as well. Now within here is I'm going to say let's get the five latest hacker news stories again and let's create a document on Google Docs leveraging Composio. Again, this is a completely different harness, but the one thing that is in common is it has that layer of all of those different tools. So you can think of a composio almost like the universal tool adapter for agents. Regardless of whether you're leveraging codeex one minute, you're leveraging open claw the next minute, claude code the next. You can know, especially with the CLI, it's really easy to have these tools be portable within here. It says, "Got it. I'm grabbing the five latest hacker news items. Then I'll use Composio to make the Google doc." And then here we go. Here are the five latest hacker news stories. It's streaming out here on Telegram for us. We have these stories that it listed out for us. And the idea of how you can stitch some of these things together is pretty remarkable because what you can do is you can effectively set up things like cron jobs potentially for your open claw and you can say hey at 9:00 I want you to go to hacker news. Give me a brief of everything that's going on. Then at a certain interval say once a day I want you to check anthropic see if they have any new blog posts. Then additionally you can also set this up in a way where you can have it go through your email and draft responses within your own voice for instance. And the really powerful thing with this is when you begin to combine all of these things within workflows. And the amazing thing that I absolutely love with this is you can create those workflows with just natural language. For instance, what you can do is you can create a workflow within a skill. You can have something like check my email every day at 8 a.m. Then check my calendar. Then go to hacker news. Give me the five latest stories. Put them within a Google doc. And then draft any emails that are needed. You have the ability to combine all of these different services in very creative and powerful ways without having to worry about the minutiae like you see here of setting up. The thing that I want to emphasize with this is if you were to set up all of these different tools yourself, this goes without saying would have taken a lot of time. Now, not to say you're going to leverage all of these, but even a dozen of these, even a couple dozen of these is as you realize your agent needs these different things. All of those different points of integration are often times of friction cuz every provider does have a little bit of a different configuration depending on the agent harness and how different things are going to be set up. So, this just gives you a one universal unified layer for having your agents access a ton of tools and be able to stitch them together in workflows of whatever you find creative. So, let me know in the comments below how you're thinking of potentially leveraging this. But otherwise, if you found this video useful, please comment, share, and subscribe.