MCP is Becoming Increasingly Trending, is the AI ​​Agent Sector Going Viral Again?

[TL；DR]:

The AI ​​Agent sector has risen by 32% in the past seven days, with its market capitalization climbing to $5.6B, and market attention has increased dramatically.

Several AI Agent protocols such as Swarms, .fun, and SKYAI have launched MCP services, but most of them are in the early stages where concepts outweigh practice, and their products are not very mature.

If MCP wants to become the “HTTP” of AI, security and compatibility must come first.

Introduction

Since last week, the crypto market has seen a long-awaited rebound, with the overall market value rebounding significantly and investor sentiment improving significantly. In this rebound, the AI ​​Agent sector has performed particularly well, with a 42.3% increase in the past seven days, and a market value of $6.1B, with market attention increasing sharply. This article explains the technical driving logic behind MCP and ACP.

AI Agent Sector Leads the Rise, and MCP Catalytic Effect Begins to Emerge

MCP is reshaping the value narrative of AI Agent and driving the recent strong rise of this sector. I have mentioned in many blog posts discussing the market downturn that the integration of AI and blockchain will be one of the important trends in the future crypto market, and the leading rise of the AI ​​sector this time undoubtedly provides strong evidence for this view. In this wave of the AI sector’s rise, MCP technology undoubtedly played the role of a catalyst.

Source: coingecko

MCP (Model Context Protocol) is designed to solve the fragmentation problem when AI models interact with external tools and data sources. In simple terms, MCP is akin to a “universal plug” for AI, enabling models to connect securely and standardly to databases, file s, or third-party APIs to obtain real-time data and perform tasks. Its emergence coincides with the critical moment when AI Agents move from “understanding” to “ution.”

In other words, the important role of MCP is that it expands the application scenarios of the AI Agent. In the past, AI models were mostly off-chain tools that interacted with the blockchain through APIs, which were inefficient and limited by centralized interfaces. Today, MCP provides AI models with on-chain identity and contextual awareness, enabling them to function as “on-chain cognitive utors.” For example, in DeFi, AI market makers can utilize MCP to analyze order book depth and cross-chain sentiment indicators in real-time, dynamically adjust liquidity scheduling, and replace traditional static models. This practicality has given the market an opportunity, and funds have begun to flow into the AI ​​sector.

Source: @Lawrence

From a broader perspective, the rebound of the cryptocurrency market and the rise of the AI ​​sector are not isolated events. The advancement of AI technology provides a foundation for on-chain applications, while the decentralized nature of blockchain provides a trust and incentive framework for AI. MCP, as a bridge, fills the technological gap and ignites the imagination of investors. Therefore, in the crypto market, which claims to be valued at the market dream rate, there is naturally a prerequisite for hyping the AI Agent sector.

MCP Application Case Analysis: Low Product Maturity and Rising Bubbles

To better understand the practical application of the MCP protocol, we share some AI Agent protocols that have adopted MCP.

Swarms: Enterprise-level multi-agent collaboration framework Swarms recently launched an MCP server for exchanges to implement token data analysis and automatic transaction ution within the production-ready swarms-rust framework. The team has now integrated optional task uation tools into the agent framework.

Source: @swarms_corp

.fun: AI Agent data platform .fun recently launched its own MCP server, providing developers and non-technical personnel with a plug-and-play agent-specific MCP server. Users can easily create and deploy their own AI agents without any complex configuration.

DeMCP: A live Web3 MCP marketplace that provides discounted access to large language models such as GPT-4 and Claude. Unlike traditional AI service access methods, DeMCP uses a trusted ution environment (TEE) and on-chain security mechanisms to ensure data security and privacy. This means that users can enjoy AI services without worrying about data leakage or abuse.

SkyAI: SkyAI, based on the BNB Chain, provides multi-chain data access and AI agent deployment. The project creatively combines MCP with zero-knowledge proof, allowing AI agents to implement smart contract auditing, token tracking, and cross-chain analysis.

Source: @SKYAIpro

We introduced this project in last week’s article, “Will BNB Chain Link the SOL Chain to Revive the On-chain Eco?” SKYAI’s presale attracted over 110,000 addresses to participate, raising a total of 83,848 BNB (approximately $50 million). After the official TGE, the market value once rose to a maximum of 76 million, and the market hype effect was obvious.

Dark: DARK is an eco designed for AI agents that integrates economic s and technical infrastructure. Its goal is to build a network infrastructure that supports anyone to create, host and profit from MCP services. It currently runs two MCPs:

• Dark Forest MCP: Dark Forest MCP gives SPUTNIK (a competitive AI agent from DARK Games) the ability to explore unknown areas in foggy maps, analyze its own coordinates, and plan planet-hopping routes.
• Solana MCP: Provides full capabilities for interacting with the Solana blockchain.

Other projects, such as OpenServ’s AWS integration, Holoworld’s Agent Studio, and ElevenLabs’ ability to order pizza through MCP and Claude, have all achieved breakthroughs in segmented scenarios.

However, most MCPs in both Web2 and Web3 are in the early stages of being more conceptual than practical, especially many AI Agent protocols that use MCP as a selling point, but the product maturity is not high. At this stage, when the concept bubble is hyped, we still need to pay attention to those protocols that can land and bring value updates.

MCP Still Has Hidden Dangers, What Other Cards Can AI Agents Play?

Although the MCP protocol has shown great potential for application and market prospects in the cryptocurrency market, it still faces numerous challenges and hidden dangers on its development path.

Firstly, the popularity of MCP is not high enough, and many large AI services have yet to integrate MCP. In particular, the support for MCP by different s and platforms is limited, which in turn restricts the further development of MCP in the crypto market.

Furthermore, the openness of MCP will also bring some privacy leakage issues. For example, 23pds of SlowMist Technology warned that MCP may have loopholes, and hackers can leak user conversation history by manipulating the AI ​​. In fact, as an open protocol, MCP must handle the interaction of multiple data sources, thereby increasing the attack surface. Hackers may indeed achieve malicious purposes by forging requests or tampering with data. For example, when an AI Agent accesses on-chain data, if the server is compromised, user data may be stolen. In addition, when users authorize AI Agent to access sensitive information, if a third party accesses it without authorization, the trend will seriously violate privacy.

Source: @trailofbits

In other words, the infrastructure of Web3 AI Agent is not perfect, and the “pits” such as the data layer and the oracle layer have not been filled. If MCP expands blindly, it may amplify the vulnerability of the . The value of technology lies in robustness. If MCP wants to become the “HTTP” of AI, security and compatibility must come first.

In the face of the trust crisis surrounding MCP, the emerging ACP (Agent Commerce Protocol) protocol offers an alternative possibility. Different from the “technical adaptation” positioning of MCP, ACP focuses on building a decentralized commercial settlement layer and introduces the “Proof of Intent” mechanism to ensure that the agent’s behavior is entirely consistent with the user’s goals.

In summary, in my opinion, the future of AI Agent may move towards a “protocol layered” architecture: MCP is responsible for technical connections, ACP handles value exchange, and the underlying layer requires infrastructure support such as distributed identity (DID) and decentralized storage (such as IPFS).

Source: @DeMCP_AI

Of course, the current trend of AI Agents and even the integration of AI and crypto is still in the process of technological evolution and multi-party discussion, and the direction is not yet clear. However, the author has always believed that only by placing users’ efficiency and privacy needs at the core of the underlying logic can greater economic value be stimulated. The author is also willing to communicate with you to discuss more insights and observations in this field.