OutSystems introduces official MCP support for both OutSystems ODC and OutSystems 11 (O11). This marks an important step forward in agentic development, something that CoolProfs has been actively working on for quite some time. But what exactly is an MCP?
What is an MCP?
An MCP (Model Context Protocol) can be seen as a standardized connection between an AI agent and an external system. Everyone is familiar with the ChatGPT interface, where AI generally generates text. Through an MCP server, however, a generative AI system can also perform actions in a system or retrieve and analyze information.
In the case of OutSystems, an AI agent gains access through the MCP protocol to an application, library, or module, enabling the agent to analyze an application or even build entirely new functionality based on a prompt.
MCP Server for ODC: The Road Ahead
With the introduction of MCP support for ODC, AI coding agents such as Claude Code, Codex, Cursor, and Kiro can collaborate directly with the OutSystems platform. From a single conversation with an AI agent, you can build applications, generate AI agents for use within an app, create workflows, or develop reusable libraries.
Another use case is uploading and publishing C# libraries when custom code is required within a low-code application. The AI agent understands ODC’s built-in capabilities and operates within the same governance framework, authorization model, and audit trail as a human developer.
OutSystems outlines an even more interesting future. Through MCP, your AI coding agent of choice can not only interact with OutSystems but also with many other tools already in use within your organization. For example, it could retrieve a user story from Jira, build the functionality, automatically deploy it to a test environment, and execute tests. AI could then send a summary of the results to a Slack channel.
This confirms the direction we have already chosen at CoolProfs. Consultants spend less time on actual coding and increasingly focus on directing, reviewing, and optimizing AI agents. The role is shifting from builder to orchestrator.
In our vision of Agentic Delivery, the intelligent use of AI agents within a controlled development process is central. Naturally, this remains a process guided and supervised by people. Why do we believe in this approach? Because the combination of our experienced consultants, AI agents, and the OutSystems platform enables applications to be developed faster while maintaining quality, security, and governance.
MCP Server for OutSystems 11
MCP support is also available for OutSystems 11. Initially, agents can analyze applications in read-only mode. At a later stage, write capabilities will be added, making it possible for AI to create screens and logic. All of this will happen through the familiar Service Studio environment used by developers. This is good news for customers who are still fully operating on O11.
At the same time, they can also benefit from ODC, since every OutSystems customer already has access to this platform through their license. Entirely new applications can be developed on ODC, while existing applications continue to run and be maintained on O11. Thanks to OutSystems’ interoperability principle, the applications can continue to work together and share data seamlessly.
Our vision
At CoolProfs, we see ODC as the strategic destination for agentic software development. This is where the most advanced capabilities for AI agents, automation, and modern software delivery are emerging. This is why we have already launched our Cool Studios: industry-specific environments where consultants and AI systems work together to deliver OutSystems applications faster.
It will come as no surprise that we use AI agents such as Claude Code and connect them to ODC’s Mentor system through MCP – we had access to the beta. Do you also have a (legacy) project that you would like to modernize into low-code with lower costs? Get in touch!
Onno Poelmeyer, Consultant, CoolProfs