MuleSoft, your way is an AI toolkit that allows developers to build, manage, and deploy Mule apps via their preferred MCP-supported Integrated Development Environment (IDE) or AI Clients. With this collection of AI tooling, users can extend our developer and generative capabilities to their development experience of choice.
Last month, MuleSoft launched a Model Context Protocol (MCP) server embedded within our development platform. We introduced a host of new capabilities focused on extending the development experience across MuleSoft – from a visual, functional, and generative perspective – to third-party AI IDEs and clients.
This unlocked the ability for our customers to build applications via the development experience of their choosing, whether that be in MuleSoft, via popular third-party AI IDEs such as Cursor and Windsurf, or via an agentic client like Claude. All of this is possible thanks to our investment in MCP and extending the visual experience of Anypoint Code Builder via our VSCode extension pack.
The announcement included a host of different features; customers are now able to use natural language to create Mule Projects and flows, deploy applications locally or to CloudHub 2.0/RTF, retrieve application details, and manage Anypoint Exchange assets.
We’re continuously working to extend even more of our development capabilities via our MCP server, in an effort to extend our AI toolkit and provide the most flexible experience for our customers, supporting a rapid composability and their broader journey to agentic transformation.
This article serves as the first in a regular series where we’ll be rounding up both new and existing functionalities across MuleSoft available via the MCP server, and by extension, your AI development experience of choice.
If you’d like to get started, be sure to download our VSCode extension pack and take a look at the MCP server ReadMe. Watch an overview demo and read all about the new features below, all of which are now generally available, wherever your AI development workflow takes you.
It’s MuleSoft, your way.
MCP server generation
For more than a decade, MuleSoft has been a champion of composability; of approaching your IT infrastructure as interchangeable building blocks that can be added to or subtracted from in such a way that makes sense for your business, making scaling IT as frictionless an experience as possible. When it comes to agentic transformation, there’s a new sheriff in town: actionability.
Actionability focuses on taking the building blocks of your composable IT architecture and making them agent ready. MCP is the standard around which the industry is settling in order to expose traditional IT applications to AI clients and agentic AI. We refer to this as agent-to-system communication, and it forms one of the foundational pillars of actionability. The capabilities outlined in this very article are a good indication as to the actionability that MCP has afforded MuleSoft, and we want to provide that same benefit to our customers.
Our customers are already able to expose their Mule apps via MCP thanks to our MCP Connector. With this release, the process is now even simpler. Users can now use natural language to expose their applications to AI clients and agents via MCP.
This is truly a game-changer for MuleSoft customers making shifting gears toward agentic transformation. They can leverage their existing MuleSoft technical investment along with the hundreds of connectors we currently offer to the most popular apps and technologies, and make them agent-actionable.
As the world orients around MCP, enabling AI to quickly and efficiently engage with your IT ecosystems is paramount. As we learned from the era of digital transformation, speed wins; by making the generation of MCP servers conversational, it becomes simple to expose your systems to AI so they can perform any number of tasks in a secure, reliable and authenticated way, freeing up Muley’s to focus on AI innovation.
The new natural language features available via the MCP server are:
- Create MCP servers using natural language processing (NLP)
- Create a basic flow
- Configure flows
- Set up hundreds of MuleSoft out-of-the-box connectors
- Start the application using local run and debug
- View available deployment options for applications
- Deploy to CloudHub 2.0
- Deploy to Runtime Fabric
API and policy management
A robust API Management (APIM) strategy is essential for navigating the evolving landscape of AI and agentic systems. As these tools gain traction across IT, we’re seeing developers increasingly building APIs specifically for agent use. A well-thought-out management approach is thus crucial to expand your API ecosystem to empower agents with the necessary level of controlled actionability.
That’s why we’re excited to announce that our customers are now able to execute some of MuleSoft’s core APIM capabilities via natural language. This includes standing up an API instance in a new application and performing critical configurations such as establishing the API credentials. Users can also apply basic authentication security policies. It’s now easier to manage and scale your API footprint with MuleSoft. Take a look at these new APIM MCP tools in the demo below.
The new natural language features available via the MCP server are:
- Add an option to set up an API instance in a new application
- Create a basic API instance
- Configure the API instance in a project
- Set up credentials for the API
- Start the application using VS Code debug
- View available security policies for the API
- Apply basic authentication to the API
- Apply a policy to the API
Generative API specifications
There’s a huge emphasis now being placed on time-to-value and on broadstroke efficiency across the software development life cycle. With so many AI tools available to accelerate the engineering process, developers can no longer afford to spend their precious time on more rudimentary coding work.
Last month, we also launched a host of new generative capabilities across MuleSoft, including the ability to create API Specifications using natural language. Generative API specifications are now available via the MCP server so that, no matter where you’re building Mule applications, you’ll never have to manually craft an API spec again. Instead, you can spend more time focusing on AI innovation and agentic transformation.
The new natural language features available via the MCP server are:
- Generate APIs that have been validated semantically and syntactically
- Modify previously generated outputs in multi-turn conversations
- Review previous generations in past chats
- Improve code quality via the Salesforce code validation layer
Platform Insights and Reuse Metrics
With so many new AI tools being onboarded and integrations being stood up to make them actionable, your IT infrastructure will probably be under more strain than usual. Given the autonomous nature of a lot of the tools being introduced in this particular technological shift, It’s critical to have clear visibility into your ecosystem.
Visibility can help inform which guardrails to place during rapid scaling, as you can understand exactly how AI is interacting with your critical IT infrastructure and ensure that agents do not exceed their remits. Understanding which systems your agents are using the most can drive investment decisions and help plan your strategic approach to AI.
That’s why we’re happy to announce that both platform insights and reuse metrics are now available via our MCP server. Platform Insights offers visibility into API, flow, and integration performance and consumption. Usage metrics track how MuleSoft assets, such as APIs and reusable apps, are adopted across teams and environments. Both tools use natural language processing for easy querying with example prompts provided for each.
The benefits of these tools include optimizing API portfolios, identifying operational hotspots, promoting best practices for API and asset reuse, and understanding platform leverage across different organizational parts.
The new natural language features available via the MCP server are:
- Obtain data on trends in flows, messages, and data throughput
- Derive insights and observations from the platform data
- Obtain information on the performance of a particular API or application
- Provide error rates, latency metrics, and call volume trends
- Help users understand traffic patterns
- Enable users to make decisions about their applications and APIs
To learn more, read Introducing Two Powerful MCP Tools: Platform Insights and Reuse Metrics.
What’s next?
That’s all for our July MCP Server announcement, but be sure to tune into the next installment where we’ll be focusing on:
- Custom metadata: Set custom metadata for operations within your application
- Test a connection: Verify whether connections are valid or invalid
- Generate a DataWeave transformation: Enable clients to generate data transformations using our proprietary AI services
- Add and run rulesets: Add a governance ruleset to your API and run the ruleset to understand whether or not your API is conformant
Want to learn more about embracing the AI future? Read our guide, From Digital to Agentic Transformation, to get started on your agentic journey.
