The more you know, the more you realize you don’t know; amateurs are like a blade of grass!
Follow us, and we will improve together! If you star us, we will have more stories to tell!
Editor: Amateur Grass
Recommended:https://t.zsxq.com/rufwd
3 Steps to Achieve Dynamic Load Balancing in Spring Boot: A Complete Guide to etcd Integration
5 Minutes to Master the Hottest AI Protocol of 2025, Complete Demo of Spring AI + MCP
Why Did Java Abandon the Observable Observer Pattern?
Explosive, DeepSeek-V3-0324 Boosts Coding Efficiency!
Understand the Stream Collector in JDK 24 in One Article! Provides Window Operations Similar to Flink!
Will the Model Context Protocol (MCP) Explode Like HTTP?
In less than half a year of development, the Model Context Protocol (MCP) has seen unprecedented growth. It even has the potential to surpass the HTTP protocol, as nowadays, no AI means no chat, and no AI means no financial reports.
Regardless of how well AI is used internally in companies, the hype around AI must be maintained to appear trendy. Otherwise, not only will financial reports look bad, but the AI hype will also fall flat, and shareholders will not be pleased.
Many people are learning AI and embracing it, but what about the MCP protocol? They may have memorized many prompts, but when it comes to AI applications, agents, function calls, etc., they find it difficult to proceed.
I have previously written articles related to the MCP protocol, but I feel it is not enough. I believe that MCP will see unprecedented development, just like the current HTTP protocol, and may become ubiquitous in the future.
Because the MCP protocol can truly facilitate the implementation of AI applications. Let’s look at the following scenarios:
- Querying AI to access existing data in the local database to assist in development
- Asking AI to search GitHub Issues to determine if a problem is a known bug
- Using AI to send feedback on a PR to a colleague’s instant messaging software (e.g., Slack) for code review
- Using AI to query or even modify current AWS or Azure configurations to complete deployment
- Using AI to obtain API meta capabilities from registration services like Nacos (it is said that the upcoming Nacos 3.0 will bring AI capabilities)
- …
These functionalities are currently becoming a reality through MCP. For example, Cursor’s MCP (<span>https://docs.cursor.com/context/model-context-protocol</span>) and Windsurf MCP have played significant roles in various applications.
Why is the implementation of AI applications progressing so slowly? There are many reasons, one of which I believe is the lack of an open, universal, and consensus-based protocol standard.
Thus, Claude (Anthropic) saw the opportunity and led the release of an open, universal, and consensus-based protocol standard, which is the common MCP protocol used by many major companies.
The best opportunity for the release of this protocol should have belonged to OpenAI. If OpenAI had promoted the protocol when it first released GPT, I believe everyone would have accepted it. However, OpenAI has become CloseAI, only releasing a closed GPT, which is a pity.
After Claude released MCP, the official Claude Desktop opened MCP functionality and promoted the open-source organization Model Context Protocol (<span>https://github.com/modelcontextprotocol</span>), with participation from various companies and communities. Below are some examples of MCP servers released by different organizations.
- Git – Read, manipulate, and search Git.
- GitHub – Repo management, file operations, and GitHub API integration.
- Google Maps – Integration with Google Maps to obtain location information.
- PostgreSQL – Read-only database queries.
- Slack – Sending and querying Slack messages.
There are many more examples of MCP, all of which can be viewed and learned from the modelcontextprotocol at <span>https://github.com/modelcontextprotocol/servers/tree/main/src/git</span>.
In addition, Grafana, JetBrains, Stripe, AWS, Atlassian, Google Calendar, Kubernetes, X (Twitter), YouTube, etc., are all embracing the open-source MCP.
I won’t say much more; I let AI predict the following, and DeepSeek said,<span>MCP has the potential to become the "HTTP protocol of the AI era"</span>. Meanwhile, industry analysts point out that the MCP tool market size may exceed $50 billion by 2026, especially in the fields of AIoT and industrial automation.
If you are still worried about not finding a job, then start learning about MCP protocol integration cases. I believe you can, and you will definitely outperform others.
Additionally, I suggest following this learning path:<span>MCP defines interaction standards -> Functions provide atomic capabilities -> Agents integrate capabilities to solve problems</span>. Just as TCP/IP and HTTP protocols support web application development, MCP is expected to become the foundational protocol for the large-scale implementation of AI agents.
After rambling on, I just want to express one thing,MCP is expected to reconstruct the AI technology stack like HTTP and become the core protocol of the agent network! Learning MCP and related ecosystem projects will surely bring good fortune!
