The Progress® OpenEdge® AI Coding Assistant is a solution designed to support Advanced Business Language (ABL) developers working within the OpenEdge ecosystem. It provides developers with intelligent, context-aware assistance for writing, reviewing, and modernizing ABL code.

OpenEdge ABL developers often encounter challenges when trying to use AI for code quality improvement. Existing large language models (LLMs) have little to no knowledge of ABL, and training or fine-tuning custom LLMs requires large datasets and specialized expertise that are not readily available. By leveraging purpose-built generative AI (GenAI) capabilities, the OpenEdge AI Coding Assistant solution streamlines development workflows and automates repetitive tasks, including:
  • Code completion
  • Review automation
  • Documentation building
  • Unit test creation
  • Code quality checks
  • Security scans
  • Continuous integration/continuous delivery (CI/CD) improvements
  • Legacy code modernization
The solution comprises the following components working together:
  • An artifical intelligence (AI)-enabled Integrated Development Environment (IDE)—This guide uses Windsurf to demonstrate AI-powered ABL development. The setup is tool-agnostic and supports Visual Studio Code (VS Code), Cursor, and other MCP-compatible environments. For more information on how to download and set up Windsurf, see Install and set up Windsurf.
  • Progress® OpenEdge® AI Assistant (formerly MCP Connector for ABL)—A VS Code extension that connects the IDE to a live ABL language reference and documentation context using the Model Context Protocol (MCP). The OpenEdge AI Assistant gives the AI agent accurate, real-time knowledge of ABL grammar and OpenEdge APIs. For more information, see Use the OpenEdge AI Assistant.
  • The OpenEdge ABL extension by RiversideSoftware—Provides language support, syntax awareness, and OpenEdge-specific tooling inside the IDE.
  • An agentic AI framework—This guide uses Cascade, the agentic AI framework built into Windsurf that understands user intent and codebase context to support code generation, error detection, and task automation. Other supported IDEs provide equivalent agentic frameworks.
  • Agentic framework features—Guide the behavior of the AI agent with ABL-specific syntax rules, project conventions, and repeatable task workflows. This guide demonstrates Cascade rules, memories, and workflows features in Windsurf.
Note: Progress also offers the OpenEdge AI Coding Assistant for ABL Accelerator Service—a professional service offering that helps you implement, customize, and scale AI-enabled workflows, including codebase indexing, compliance enforcement, and modernization initiatives such as refactoring for OpenEdge 12.8 and Progress Application Server (PAS) for OpenEdge. This guide covers the self-configured product components. For information about the service offering, contact your Progress representative.

How to set up the OpenEdge AI Assistant solution

To set up the full solution, follow these steps:

Step 1: Install and set up Windsurf

Step 2: Use the OpenEdge AI Assistant

Step 3: Configure the OpenEdge ABL development environment

Step 4: Configure and manage Cascade for advanced AI-assisted workflows

How the OpenEdge AI Coding Assistant solution works

When you type a prompt in Cascade, for example:
Create a class that reads customer records from the Sports2020 database
The solution coordinates several components behind the scenes to deliver accurate, ABL-aware results.

Here is what happens:

  1. Cascade receives your prompt—Cascade, the agentic AI framework in Windsurf, interprets your intent based on your prompt, your open files, and any rules or memories you have configured.

  2. Cascade queries the OpenEdge AI Assistant—Before generating code, Cascade calls the OpenEdge AI Assistant using the Model Context Protocol (MCP). This is a standardized communication protocol that allows Cascade to request specific ABL grammar rules, language reference entries, or documentation context at runtime.

  3. The OpenEdge AI Assistant retrieves relevant context—The OpenEdge AI Assistant (MCP server) retrieves relevant OpenEdge documentation, syntax definitions, and rules from a vector database using Retrieval-Augmented Generation (RAG). This context is specific to your query. For example, the correct syntax for DEFINE TEMP-TABLE, buffer handling conventions, or database access patterns.

  4. The context is passed to the LLM—Cascade augments your original prompt with the retrieved ABL context and sends everything to the configured LLM, such as ChatGPT, Claude, or another supported model.

  5. The LLM generates an ABL-aware response—Because the model receives accurate ABL documentation alongside your prompt, it produces code that follows correct OpenEdge syntax and conventions rather than hallucinating generic patterns.

  6. Cascade applies the result—Cascade writes the generated code directly into your workspace, applies your configured rules, for example, using VAR instead of DEFINE VARIABLE, and flags any compilation issues detected by the ABL language server.

This entire cycle happens in seconds, with no manual lookup required.

Key features

The OpenEdge AI Coding Assistant solution provides the following key features:

  • IDE integration and tool-agnostic setup—Supports integration with multiple IDEs to help developers stay in flow and benefit from AI-driven workflows and rule-based enhancements without needing to switch tools.
  • OpenEdge-specific AI code suggestions—Provides AI-powered code suggestions tailored to OpenEdge syntax and business logic.
  • Layered rule enforcement—Enforces layered rules by incorporating customer-specific, framework, and ABL requirements, ensuring that code suggestions are both relevant and high-quality.
  • Supports onboarding—Developers new to ABL get accurate guidance without manual documentation searches.
  • Flexible LLM selection—Provides multi-LLM support to help developers choose between models like ChatGPT, Claude Sonnet, and more.
  • Linting and modernization—Provides real-time linting by analyzing source code to detect errors, bugs, stylistic issues, or deviations from coding standards. It also offers modernization support for upgrading to OpenEdge 12, thereby improving ABL code maintainability, performance, and security.
  • Context-aware assistance—Provides context-aware assistance using RAG, which is an AI architecture that enhances the output of LLMs by retrieving relevant information from external sources, such as internal documentation, databases, or knowledge graphs, before generating a response. OpenEdge AI Coding Assistant grounds AI responses in your actual codebase, documentation, and rules for more accurate, context-aware, and verifiable responses.
  • Use of MCP—Ensures standardized, extensible integration of the OpenEdge AI Coding Assistant solution across IDEs and development environments.

Unlike generic coding assistants, this solution is designed specifically for the unique patterns and ABL coding practices of OpenEdge.