# Why MCP is dead & How I vibe now

## Метаданные

- **Канал:** AI Jason
- **YouTube:** https://www.youtube.com/watch?v=fG95XsBO5U4
- **Дата:** 24.01.2026
- **Длительность:** 10:51
- **Просмотры:** 19,017

## Описание

25 prompts to unlock true Claude cowork power: https://clickhubspot.com/3864a6

🔗 Links
- Try Superdesign Claude Code Design OS Skill: https://x.com/jasonzhou1993/status/2014270488391164012
- Get the UI design prompt: https://www.aibuilderclub.com/
- MCPorter: https://github.com/steipete/mcporter
- Follow me on twitter: https://twitter.com/jasonzhou1993

⏱️ Timestamps
0:00 The problem
2:56 CLI + Skill
4:38 agent browser example
6:53 MCPorter
8:33 Add new mcp skill
9:34 Superdesign CLI skill

## Содержание

### [0:00](https://www.youtube.com/watch?v=fG95XsBO5U4) The problem

Many people didn't know you can actually use simple open source tools to reduce more than 70% of token consumptions for your coding agents and extend it capability in much more scalable way. So every MCP server you add comes with this bundle of different tools and each tool comes with this JSON schema and all those information is part of context no matter whether the task that agent is doing at moment is relevant or not. So it just unnecessary eats up loads of context window. A much more scalable and easy to use method in my opinion is this combination of skills plus CRI tool. It allows you extend your agent capability to hundreds different integrations with only minimum amount of token consumptions. In my experimentation, some of them can lead to more than 70% of token consumption reductions. And this is what I want to take you through today with step-by-step process of how you can do it and special agent skills to enable that. But before we dive into that, entropy released cloud code works which is special version of cloud code that aim to automate not just coding but knowledge works. It gave us sneak peek of how white collar work going to be done in future. But many of us are still probably early stage experimenting it or figuring out what's the best way to unleash it full power. That's why I want to introduce you to this free research and material HubSpot provided. They did a whole bunch of experimentations across all the capabilities cloud co-work is providing and test across hundreds of use cases in real world. So they figure out 12 advanced prompts and use case that is programmatically useful across things like automatically analyze content performance data and propose new trends to create common business documents like job description SOP email sequence as well as getting cloud co-work continuously organize your file system for business and even automated competitive intelligence reports that can be scheduled every week and each one comes with a readytouse prompt that will trigger a predefined cloud co-work process. So if you're interested in understanding what the real paradigm shift cloud co-work is going to provide, I highly recommend you go take a look. I put a link in the description below for you to download this resource for free and thanks HubSpot for sponsoring this video. Now let's get back to how can you use skill plus CRI to replace MCPS. So agent skill concept has been introduced end of last year but it start getting extremely popular past few weeks and the concept is simple. You basically create a skill. m MD which including a snippet of prompt that it will be injected to the agent when the skill is retrieved plus the list of resource the agent can use to complete the tasks. Every time when you load up a skill it only add a minimum amount of token in the context window. Basically cloud code only has this one tool called SQL and inside this two description it has a section of available skills. So you can see that every single skill you add only additional 10 to 50 tokens. That means with the amount of token that is taken from MCP tools in this screenshot, you can easily add about 4,000 different SKUs, which is more than anything you

### [2:56](https://www.youtube.com/watch?v=fG95XsBO5U4&t=176s) CLI + Skill

will possibly need. So when agent execute skill, the prompt you load up here will be added to the context. And based on those context, you can include instruction about how to utilize the rest of resource to get either additional relevant context or execute some predefined scripts. And we already see a whole bunch of developers moving towards the skew based setup that is creating a much more token efficient agent. One very popular example is actually manis. So manis actually did a whole bunch of knowledge sharing about how they do context engineering and one of the key things they did is that they break down all the agent tools into different levels. Their fundamental tools that is always useful like read, write, edit file capability, things like that will be always loaded as normal tools. But for the rest of less frequent tools, they're most likely put under a command line that agent can run to execute. For example, if you add an MCP to Manis, the MCP is never exposed to Manis. Instead, Manis agent has this manus MCP CRI tool the agent can call to execute MCP and they just add a little prompt snip to tell agent what type of MCP it can run. So that when agent need to use this tool, it can just run the command line to run MCP instead. And because it is CRI, it can also run other useful commands to get more detail and debug if there's any error. And because it's CRI, agent out ofbox can do more complicated actions like piping multiple different actions together in one go with some default system function like wait for a certain amount of time and even do more complicated action like passing result A to result B long as it understand the output schema well. And this is just such a more powerful and token efficient way to extend agent's capability. And to showcase why this way

### [4:38](https://www.youtube.com/watch?v=fG95XsBO5U4&t=278s) agent browser example

better, I just quickly go through this one example. So we all want cloud code to be able to use a browser test application or review design. The browser tool like playride or even cloud code's own chrome mcp is very complicated because it will load up a whole bunch of different tools for agent to open a tab click on something drag something typing something and that can easily take at least 2% of context window which in this case left only 125,000 effective context window that agent can use like in this example I use chrome dev2p to do the UI test of super design platform and after finish a basic UI has like this it consume a bunch of token and in the end it has 87,000 token left but on the other side with agent browser what they did is that they didn't introduce a MCP tools instead they build a CRI package which is command that both you and agent can run to do all sorts of different browser actions so it can just run those command line take actions and get context back and it did a lot of different optimizations to making sure the information it return really help agent understand the page structure what kind of interactive elements are. So the content it return here is also a lot more token efficient compared with other browser MCP tools and more importantly because this is CRI there it can be run directly in terminal all we need to do is just adding this agent browser skew that clearly list out how to use this CRI tool. So with this approach the effective context window you will get is already bigger than what you will get otherwise with MCP tools and all agent need to do just load up this context and start running those command line and get result back and with this specific example using agent browser with the same task when it finish it got 117,000 token left that is 70% less token consumptions and as I mentioned before this bash plus CRI tool just provides so much more freedom about how agent can execute tasks. So one thing I think really everyone should be doing is migrate off your MCP to a skill instead. But to do that you need a way to run MCP tools through CRI and this is where I found this open-source MCP CRI called

### [6:53](https://www.youtube.com/watch?v=fG95XsBO5U4&t=413s) MCPorter

MCP portter. It allows you to run MCP directly through command line and it has all the helper function the agent need to get more information about tools within this MCP server like specific schema that you need to use. So this basically allow us to do what manus has manually implemented on their site but for your coding agent and this is exactly what I did here. So instead of having also different MCPS for each MCP I will just add one skew file. For example, for context 7 MCPS, I can directly run this MPX MC porter called context 7 resolve library ID which is specific function inside contact 7 MCP server and passing on the parameters like this. And if I run this results will be directly returned back in a markdown format. And instead of adding the context 7 MCP tool, I can just add this one skill. md file for contact 7 where I put in the description about when to retrieve the skills as well as detailed description about available tools it can call. That's why my cloud code doesn't have any MCP to added. Instead, it just consume a few hundreds token for all different skills that I've added here. And I'm sure I can actually reduce that token even further by simplify descriptions. But it's already much better and can achieve all the critical actions that we need. Like if I prompted help me find the latest docu AI SDK using context 7, it will automatically load up the skill and start executing those different command line to run the MCP. So you can start immediately save whole bunch of token by converting your MCP into a skew together with MC Porter. And I've already started this process of converting all my MCP over to SKUs. I even created this one

### [8:33](https://www.youtube.com/watch?v=fG95XsBO5U4&t=513s) Add new mcp skill

skill called add new MCPS where I have this instruction to tell it how to use this MCP portter how it should install test MCP server and create a relevant skills for example I can just add any MCP paste in cloud code and tell it help me add a new MCP skill then it will load up those contexts start adding the MCP to config and check whether the MCP has been installed properly create a folder for Excel skill which detailing all the functions that agent can call to use this Excel MCP. The best thing is after adding this new skills, it only add a few new tokens into the contact window. So this example of how you can get all extended capability from MCP into skills. So you can extend agents capability much further without worrying about the performance and agent context window blow up. If you haven't done this, I highly recommend you go and have a try. If you're interested, I also put a full instruction of the add new MCP skill in AI builder club. So you can just copy and use. I have put the link in the description below for you to join yesterday and I adopt a similar methods

### [9:34](https://www.youtube.com/watch?v=fG95XsBO5U4&t=574s) Superdesign CLI skill

in a recent integration that I'm building between super design and cloud code. So many of you probably know I am building this platform called super design which is agent that allow you to vive design many different options for your products and we recently built this super design skills which allow cloud code to directly calling specialized design agent and review a design in infinite camera. In this case, we didn't go for MCP tool. Instead, we just build a CRI that cloud code can call by itself and still doing super powerful things to add a new design to this Canva and retrieve different design prompts from the prompt library to enhance the design output. Setup is extremely simple. You just need to install the package by doing mpm install-g superdesign/ci at the latest and then call superdesign login which will pop up this screen to allow you to authorize and in the end you can do mpx skills at superdesign dev/s superdesign skew so that you can add to any coding agent that you're using as a skew all we need to do just/s superdesign and prompt it load up the instructions contextually and start experimenting with different design directions all those complicated functions are done just through the skew and CRI itself. So I highly recommend you go and have a check. I hope you enjoy this video.

---
*Источник: https://ekstraktznaniy.ru/video/11433*