Infinity Build MCP fundamentals and governance
Modern enterprise engineering teams use AI-powered development environments such as Claude Code, GitHub Copilot CLI, OpenAI Codex, Devin, and custom enterprise agentic integrated development environments (IDEs). These environments typically work with text-based source code and local file structures in programming languages such as Python, TypeScript, and Java. In contrast, low-code development in Pega Platform™ uses a governed, metadata-driven rules engine with inheritance, class structures, security models, and versioning.
Pega Infinity™ Build MCP connects these development approaches by using the Model Context Protocol (MCP) to make Pega authoring capabilities available through an MCP server. Instead of manually translating architectural plans or user interface designs into visual rules in Pega Infinity Studio™, developers can use local AI coding agents to analyze relevant context and initiate structured rule-authoring operations in Pega Platform through secured APIs.
Pega Infinity Authoring plugin combines the Build MCP with Pega Infinity skills. The plugin connects external AI agents to Pega authoring tools, while the skills provide Pega-specific knowledge and best practices. A local MCP service brokers interactions with the Pega application, which executes governed authoring operations. For more information about Pega Infinity Authoring plugin architecture, see External AI agents for Pega application development
How a typical Build MCP works
When a developer asks an AI coding assistant to create or modify a Pega artifact, the AI model analyzes the request and determines the required actions. The model returns the appropriate tool calls to the local MCP server. The MCP server then invokes the relevant rule create, read, update, and delete (CRUD) or Digital Experience (DX) APIs. Pega creates the requested changes within a governed Change Request. The developer reviews and approves the changes before committing them.
The AI model does not write directly to Pega. Instead, the AI model determines the required actions, the MCP server invokes the appropriate tools, and Pega applies the changes through its standard governance controls.
The following diagram illustrates the Build MCP interaction flow.
Build MCP includes governance controls that support safe and manageable AI-assisted development. AI-generated rule changes are written only to Branch Rulesets, which prevents direct updates to base or production rulesets. MCP tools cannot merge branches or delete or withdraw Rules. Every action runs with the permissions of the signed-in Pega operator. Bundled Pega skills also guide agents to follow the Change Request workflow and obtain approval before committing changes. These controls support human review, security, and governance throughout the authoring process.
Build MCP supports AI-assisted authoring while keeping changes branch-scoped, permission-controlled, and subject to human review and approval before the changes are promoted.
Security and data boundaries
One of the most important questions for enterprise architects is “What data leaves the enterprise boundary?"
Build MCP addresses this concern by clearly distinguishing among data that remains on the developer workstation, data that is sent to the model provider, and data that is transmitted to the Pega environment.
Local configuration files, OAuth credentials, MCP process memory, and communication between the coding agent and the local MCP server remain on the developer workstation. Approved APIs support authenticated interactions with Pega. The coding agent can send the natural-language prompt and selected application context to the model provider for reasoning and response generation. These boundaries clarify which information moves between components and where that information is processed.
Organizations can use AI-assisted development while maintaining governance and visibility into data movement throughout the solution.
Relationship to other MCP capabilities
Model Context Protocol includes Build MCP as part of the broader Pega MCP ecosystem.
Build MCP focuses on development-time activities that help developers and architects build, modify, and improve applications through AI-assisted authoring.
Connect MCP focuses on integration. In this model, Pega acts as an MCP client that connects to external MCP servers to consume their tools and capabilities.
Run MCP focuses on runtime business processing. It exposes Pega workflows and business capabilities as MCP tools that external AI agents can invoke. In this model, Pega acts as an MCP server.
A Lead System Architect (LSA) should view these as three complementary capabilities:
- Build MCP: Build and modify applications.
- Connect MCP: Consume external tools and capabilities.
- Run MCP: Expose governed enterprise capabilities.
Together, these capabilities support an AI-enabled application lifecycle that a Lead System Architect (LSA) views as complimentary.
LSA perspective on Build MCP
Build MCP is more than a productivity feature. It introduces a different approach to enterprise application development by making AI an active participant in the development process.
Build MCP supports several enterprise goals that are central to the LSA role. It can accelerate development, reduce onboarding effort, promote consistency across distributed teams, and improve delivery efficiency while maintaining governance. Pega Platform governs application logic, workflows, compliance controls, decisioning, security policies, and auditability. Organizations can therefore adopt AI-assisted development while maintaining enterprise governance and control.
Check your knowledge with the following interaction:
This Topic is available in the following Module:
If you are having problems with your training, please review the Pega Academy Support FAQs.
Want to help us improve this content?