MCP in Follow | In direction of Knowledge Science

Written along with Ilan Strauss, Isobel Moure, and Tim O’Reilly as a part of the AI Disclosures Mission. Initially printed on our weblog: Asimov’s Addendum.

1. of MCP

Anthropic’s Mannequin Context Protocol (MCP) was launched in November 2024 as a method to make instruments and platforms model-agnostic. MCP works by defining servers and purchasers. MCP servers are native or distant finish factors the place instruments and sources are outlined. For instance, GitHub launched an MCP server that permits LLMs to each learn from and write to GitHub. MCP purchasers are the connection from an AI software to MCP servers—they permit an LLM to work together with context and instruments from completely different servers. An instance of an MCP shopper is Claude Desktop, which permits the Claude fashions to work together with hundreds of MCP servers.

In a comparatively quick time, MCP has turn into the spine of tons of of AI pipelines and functions. Main gamers like Anthropic and OpenAI have constructed it into their merchandise. Developer instruments similar to Cursor (a coding-focused textual content editor or IDE) and productiveness apps like Raycast additionally use MCP. Moreover, hundreds of builders use it to combine AI fashions and entry exterior instruments and information with out having to construct a whole ecosystem from scratch.

In earlier work printed with AI Frontiers, we argued that MCP can act as a terrific unbundler of “context”—the info that helps AI functions present extra related solutions to customers. In doing so, it could assist decentralize AI markets. We argued that, for MCP to really obtain its objectives, it requires assist from:

1. Open APIs: In order that MCP functions can entry third-party instruments for agentic use (write actions) and context (learn)
2. Fluid reminiscence: Interoperable LLM reminiscence requirements, accessed by way of MCP-like open protocols, in order that the reminiscence context accrued at OpenAI and different main builders doesn’t get caught there, stopping downstream innovation

We increase upon these two factors in a latest coverage word, for these seeking to dig deeper.

Extra typically, we argue that protocols, like MCP, are literally foundational “guidelines of the highway” for AI markets, whereby open disclosure and communication requirements are constructed into the community itself, fairly than imposed after the actual fact by regulators. Protocols are basically market-shaping units, architecting markets via the permissions, guidelines, and interoperability of the community itself. They’ll have a huge impact on how the industrial markets constructed on high of them perform too.

1.1 However how is the MCP ecosystem evolving?

But we don’t have a transparent thought of the form of the MCP ecosystem at the moment. What are the most typical use instances of MCP? What kind of entry is being given by MCP servers and utilized by MCP purchasers? Is the info accessed by way of MCP “read-only” for context, or does it permit brokers to “write” and work together with it—for instance, by modifying recordsdata or sending emails?

To start answering these questions, we have a look at the instruments and context which AI brokers use by way of MCP servers. This offers us a clue about what’s being constructed and what’s getting consideration. On this article, we don’t analyze MCP purchasers—the functions that use MCP servers. We as a substitute restrict our evaluation to what MCP servers are making obtainable for constructing.

We assembled a big dataset of MCP servers (n = 2,874), scraped from Pulse. We then enriched it with GitHub star-count information on every server. On GitHub, stars are just like Fb “likes,” and builders use them to point out appreciation, bookmark tasks, or point out utilization.

In observe, whereas there have been loads of MCP servers, we discovered that the highest few garnered many of the consideration and, seemingly by extension, many of the use. Simply the highest 10 servers had practically half of all GitHub stars given to MCP servers.

A few of our takeaways are:

1. MCP utilization seems to be pretty concentrated. Because of this, if left unchecked, a small variety of servers and (by extension) APIs may have outsize management over the MCP ecosystem being created.
2. MCP use (instruments and information being accessed) is dominated by simply three classes: Database & Search (RAG), Laptop & Internet Automation, and Software program Engineering. Collectively, they acquired practically three-quarters (72.6%) of all stars on GitHub (which we proxy for utilization).
3. Most MCP servers assist each learn (entry context) and write (change context) operations, exhibiting that builders need their brokers to have the ability to act on context, not simply eat it.

2. Findings

To begin with, we analyzed the MCP ecosystem for focus threat.

2.1 MCP server use is concentrated

We discovered that MCP utilization is concentrated amongst a number of key MCP servers, judged by the variety of GitHub stars every repo acquired.

Regardless of there being hundreds of MCP servers, the highest 10 servers make up practically half (45.7%) of all GitHub stars given to MCP servers (pie chart beneath) and the highest 10% of servers make up 88.3% of all GitHub stars (not proven).

Because of this nearly all of real-world MCP customers are seemingly counting on the identical few companies made obtainable by way of a handful of APIs. This focus seemingly stems from community results and sensible utility: All builders gravitate towards servers that remedy common issues like net looking, database entry, and integration with extensively used platforms like GitHub, Figma, and Blender. This focus sample appears typical of developer-tool ecosystems. A number of well-executed, broadly relevant options are likely to dominate. In the meantime, extra specialised instruments occupy smaller niches.

2.2 The highest 10 MCP servers actually matter

Subsequent, the highest 10 MCP servers are proven within the desk beneath, together with their star depend and what they do.

Among the many high 10 MCP servers, GitHub, Repomix, Context7, and Framelink are constructed to help with software program improvement: Context7 and Repomix by gathering context, GitHub by permitting brokers to work together with tasks, and Framelink by passing on the design specs from Figma on to the mannequin. The Blender server permits brokers to create 3D fashions of something, utilizing the favored open supply Blender software. Lastly, Activepieces and MindsDB join the agent to a number of APIs with one standardized interface: in MindsDB’s case, primarily to learn information from databases, and in Activepieces to automate companies.

The dominance of agentic looking, within the type of Browser Use (61,000 stars) and Playwright MCP (18,425 stars), stands out. This displays the basic want for AI programs to work together with net content material. These instruments permit AI to navigate web sites, click on buttons, fill out types, and extract information identical to a human would. Agentic looking has surged, though it’s far much less token-efficient than calling an API. Shopping brokers typically must wade via a number of pages of boilerplate to extract slivers of information a single API request may return. As a result of many companies lack usable APIs or tightly gate them, browser-based brokers are sometimes the only—generally the one—method to combine, underscoring the boundaries of at the moment’s APIs.

A few of the high servers are unofficial. Each the Framelink and Blender MCP are servers that work together with only a single software, however they’re each “unofficial” merchandise. Because of this they aren’t formally endorsed by the builders of the applying they’re integrating with—those that personal the underlying service or API (e.g., GitHub, Slack, Google). As an alternative, they’re constructed by impartial builders who create a bridge between an AI shopper and a service—typically by reverse-engineering APIs, wrapping unofficial SDKs, or utilizing browser automation to imitate consumer interactions.

It’s wholesome that third-party builders can construct their very own MCP servers, since this openness encourages innovation. Nevertheless it additionally introduces an middleman layer between the consumer and the API, which brings dangers round belief, verification, and even potential abuse. With open supply native servers, the code is clear and will be vetted. Against this, distant third-party servers are more durable to audit, since customers should belief code they will’t simply examine.

At a deeper degree, the repos that at the moment dominate MCP servers spotlight three encouraging information concerning the MCP ecosystem:

1. First, a number of distinguished MCP servers assist a number of third-party companies for his or her performance. MindsDB and Activepieces function gateways to a number of (typically competing) service suppliers via a single server. MindsDB permits builders to question completely different databases like PostgreSQL, MongoDB, and MySQL via a single interface, whereas Taskmaster permits the agent to delegate duties to a variety of AI fashions from OpenAI, Anthropic, and Google, all with out altering servers.
2. Second, agentic looking MCP servers are getting used to get round doubtlessly restrictive APIs. As famous above, Browser Use and Playwright entry web companies via an online browser, serving to to bypass API restrictions, however they as a substitute run up towards anti-bot protections. This circumvents the restrictions that APIs can impose on what builders are capable of construct.
3. Third, some MCP servers do their processing on the developer’s laptop (regionally), making them much less depending on a vendor sustaining API entry. Some MCP servers examined right here can run completely on a neighborhood laptop with out sending information to the cloud—which means that no gatekeeper has the ability to chop you off. Of the ten MCP servers examined above, solely Framelink, Context7, and GitHub depend on only a single cloud-only API dependency that may’t be run regionally end-to-end in your machine. Blender and Repomix are fully open supply and don’t require any web entry to work, whereas MindsDB, Browser Use, and Activepieces have native open supply implementations.

2.3 The three classes that dominate MCP use

Subsequent, we grouped MCP servers into completely different classes primarily based on their performance.

Once we analyzed what sorts of servers are hottest, we discovered that three dominated: Laptop & Internet Automation (24.8%), Software program Engineering (24.7%), and Database & Search (23.1%).

Widespread use of Software program Engineering (24.7%) MCP servers aligns with Anthropic’s financial index, which discovered that an outsize portion of AI interactions had been associated to software program improvement.

The recognition of each Laptop & Internet Automation (24.8%) and Database & Search (23.1%) additionally is smart. Earlier than the appearance of MCP, net scraping and database search had been extremely built-in functions throughout platforms like ChatGPT, Perplexity, and Gemini. With MCP, nevertheless, customers can now entry that very same search performance and join their brokers to any database with minimal effort. In different phrases, MCP’s unbundling impact is very seen right here.

2.4 Brokers work together with their environments

Lastly, we analyzed the capabilities of those servers: Are they permitting AI functions simply to entry information and instruments (learn), or as a substitute do agentic operations with them (write)?

Throughout all however two of the MCP server classes checked out, the most well-liked MCP servers supported each studying (entry context) and writing (agentic) operations—proven in turquoise. The prevalence of servers with mixed learn and write entry means that brokers should not being constructed simply to reply questions primarily based on information but additionally to take motion and work together with companies on a consumer’s behalf.

The 2 exceptions are Database & Search (RAG) and Finance MCP servers, through which read-only entry is a standard permission given. That is seemingly as a result of information integrity is essential to making sure reliability.

3. The Significance of A number of Entry Factors

A number of implications of our evaluation will be drawn out at this preliminary stage.

First, concentrated MCP server use compounds the dangers of API entry being restricted. As we mentioned in “Protocols and Energy,” MCP stays constrained by “what a selected service (similar to GitHub or Slack) occurs to show via its API.” A number of highly effective digital service suppliers have the ability to close down entry to their servers.

One necessary hedge towards API gatekeeping is that lots of the high servers strive to not depend on a single presentr. As well as, the next two safeguards are related:

• They provide native processing of information on a consumer’s machine every time attainable, as a substitute of sending the info for processing to a third-party server. Native processing ensures that performance can’t be restricted.
• If operating a service regionally is just not attainable (e.g., e mail or net search), the server ought to nonetheless assist a number of avenues of getting on the wanted context via competing APIs. For instance, MindsDB features as a gateway to a number of information sources, so as a substitute of counting on only one database to learn and write information, it goes to nice lengths to assist a number of databases in a single unified interface, basically making the backend instruments interchangeable.

Second, our evaluation factors to the truth that present restrictive API entry insurance policies should not sustainable. Internet scraping and bots, accessed by way of MCP servers, are most likely getting used (at the very least partly) to avoid overly restrictive API entry, complicating the more and more frequent observe of banning bots. Even OpenAI is coloring outdoors the API traces, utilizing a third-party service to entry Google Search’s outcomes via net scraping, thereby circumventing its restrictive API.

Increasing structured API entry in a significant approach is important. This ensures that official AI automation runs via secure, documented finish factors. In any other case, builders resort to brittle browser automation the place privateness and authorization haven’t been correctly addressed. Regulatory steering may push the market on this route, as with open banking within the US.

Lastly, encouraging larger transparency and disclosure may assist establish the place the bottlenecks within the MCP ecosystem are.

• Builders working in style MCP servers (above a sure utilization threshold) or offering APIs utilized by high servers ought to report utilization statistics, entry denials, and rate-limiting insurance policies. This information would assist regulators establish rising bottlenecks earlier than they turn into entrenched. GitHub may facilitate this by encouraging these disclosures, for instance.
• Moreover, MCP servers above sure utilization thresholds ought to clearly listing their dependencies on exterior APIs and what fallback choices exist if the first APIs turn into unavailable. This isn’t solely useful in figuring out the market construction, but additionally important data for safety and robustness for downstream functions.

The purpose is to not get rid of all focus within the community however to make sure that the MCP ecosystem stays contestable, with a number of viable paths for innovation and consumer selection. By addressing each technical structure and market dynamics, these urged tweaks may assist MCP obtain its potential as a democratizing drive in AI improvement, fairly than merely shifting bottlenecks from one layer to a different.

Appendix

Dataset

For this evaluation, we categorized MCP servers discovered on PulseMCP.com into considered one of 15 classes utilizing GPT-5 mini. We then human-reviewed and edited the highest 50 servers that make up round 70% of the overall star depend in our dataset.

The total dataset, together with descriptions of the classes, will be discovered right here (constructed by Sruly Rosenblat):

https://huggingface.co/datasets/sruly/MCP-In-Follow

Limitations

There are a couple of limitations to our preliminary analysis:

• GitHub stars aren’t a measure of obtain counts and even essentially a repo’s reputation.
• Solely the identify and outline had been used when categorizing repos with the LLM.
• Categorization was topic to each human and AI errors and plenty of servers would seemingly match into a number of classes.
• We solely used the PulseMCP listing for our dataset; different lists had completely different servers (e.g., Browser Use isn’t on mcpmarket.com).
• We excluded some repos from our evaluation, similar to those who had a number of servers per code repository and people we weren’t capable of fetch the star depend for. We additionally solely checked out servers listed on PulseMCP. Our server listing is non exhaustive.

MCP Server Use Over Time