JetBrains MPS Opens Up to AI Coding Assistants

JetBrains' specialized development tool, MPS, now integrates with AI coding agents, opening up new possibilities for AI-assisted software creation.

JetBrains has unveiled a release candidate for its upcoming MPS 2026.1 update, giving developers an early look at its new capabilities. The standout feature is the introduction of the Model Context Protocol (MCP), a new system designed to connect the MPS environment with external AI coding agents. MPS, which stands for Meta-Programming System, is a specialized tool used to create custom, domain-specific languages (DSLs). These are programming languages built for very specific tasks, like insurance policy rules or scientific modeling. Because MPS uses a unique, structure-based editor instead of plain text, standard AI assistants like GitHub Copilot have struggled to understand and assist with its projects. The new MCP, included as a bundled plugin, solves this problem by providing a way for AI agents to access the necessary context about the code, its structure, and its meaning. This allows them to offer relevant help directly within the tool.

This integration is a significant step forward for the niche but powerful community of MPS users. Developers who build and maintain complex systems with DSLs can now harness the productivity gains offered by modern AI assistants. This means tasks like generating boilerplate code, suggesting solutions to complex problems, and even learning the intricacies of a custom language can be accelerated. For businesses and CTOs, this update could lower the learning curve for new engineers joining projects that use MPS, potentially speeding up development cycles and reducing costs. It also demonstrates a critical trend in the software industry: even highly specialized, non-traditional development tools are adapting to the AI era. By bridging the gap between its structured environment and the text-based world of large language models, JetBrains is ensuring that developers using its advanced tools are not left behind in the shift toward AI-assisted programming.