Hopper

Product Introduction

1. Definition: Hopper is an agentic development environment (ADE) and integrated development environment (IDE) specifically engineered for IBM z/OS mainframe systems. It functions as a modern desktop application that integrates a genuine TN3270 terminal emulator with AI-powered automation agents capable of operating within mainframe workflows.
2. Core Value Proposition: Hopper exists to dramatically accelerate and simplify mainframe application development, JCL debugging, and system operations by introducing AI-driven automation to the traditionally manual and complex z/OS environment. Its primary value is enabling developers to interact with mainframes using high-level prompts instead of memorizing ISPF commands and navigating SDSF/JES spool output manually.

Main Features

1. AI Agent for z/OS Workflows: The core AI agent understands z/OS panel IDs (ISPF), allowing it to navigate ISPF menus autonomously. It can inspect datasets, read and edit PDS members with correct record formats, write column-strict JCL, submit jobs, and parse JES output. It specifically decodes JESMSGLG, JESYSMSG, and SYSUDUMP data into structured diagnostics like abend codes, failing steps, and source line references.
2. Mainframe-Aware Visual Panels: Hopper provides graphical, IDE-style panels for key mainframe components like datasets, jobs, PDS members, and spool output. These panels present information in a structured, searchable format, moving beyond raw 3270 data streams to a modern developer experience.
3. Integrated Real TN3270 Terminal: For direct control or legacy operations, Hopper includes a fully functional TN3270 terminal emulator with support for Program Function (PF) keys, Program Attention (PA) keys, and standard attention-key sequences, ensuring compatibility with all mainframe applications.
4. VSAM Query via SQL: The agent can query VSAM (Virtual Storage Access Method) datasets using SQL-like syntax, abstracting the complex VSAM access method commands and providing a more intuitive interface for data retrieval and analysis.
5. Automated Deployment Workflow: Hopper can orchestrate multi-step processes such as compiling code, testing via JCL, and deploying changes. This includes actions like issuing a NEWCOPY command to refresh CICS programs, with the agent pausing for user approval before executing each change to ensure safety and control.

Problems Solved

1. Pain Point: The steep learning curve and manual, time-intensive nature of mainframe development and operations. Developers spend excessive time debugging JCL errors by sifting through dense SDSF output, manually navigating ISPF panels, and writing error-prone, column-sensitive JCL.
2. Target Audience: Mainframe Application Developers, Systems Programmers, DevOps Engineers working on z/OS, and Modernization Teams seeking to improve mainframe productivity. It also serves new developers onboarding onto mainframe platforms who need a more intuitive interface.
3. Use Cases:
   • Rapid JCL Debugging: An abended job that would require an hour of SDSF triage can be diagnosed in seconds by having the Hopper agent parse the spool output and identify the abend code, failing step, and relevant message.
   • Dataset and Member Management: Quickly inspecting the contents of a PDS, viewing a specific member, or editing COBOL copybooks without manually typing ISPF commands (e.g., 3.4, =x).
   • Automated Job Submission and Monitoring: Writing a new JCL script for a batch job, submitting it, and having the agent monitor the JES return code and explain the results.
   • VSAM Data Analysis: Running ad-hoc queries against a VSAM KSDS or ESDS file using simple SQL statements instead of writing complex COBOL or REXX programs.

Unique Advantages

1. Differentiation: Unlike traditional TN3270 emulators (e.g., IBM Personal Communications, Rocket BlueZone) which are purely manual terminals, Hopper adds an intelligent automation layer. It is distinct from general-purpose AI coding assistants (e.g., GitHub Copilot) because it possesses deep, native understanding of z/OS structures (panels, datasets, JCL, JES), allowing it to operate within the mainframe environment, not just generate code snippets.
2. Key Innovation: The "agentic" capability—the AI's ability to take autonomous, sequenced actions across different z/OS subsystems (TSO/E, ISPF, JES, CICS) based on a high-level developer goal. This represents a shift from mainframe automation via scripting (REXX/CLIST) to goal-oriented, conversational interaction.

Frequently Asked Questions (FAQ)

1. What is Hopper for mainframe? Hopper is an agentic development environment (ADE) that acts as a modern IDE for IBM z/OS mainframes. It combines a real TN3270 terminal with an AI agent that can automate tasks like JCL debugging, dataset inspection, job submission, and VSAM querying, significantly speeding up mainframe development workflows.

2. How does Hopper's AI agent work with ISPF? The Hopper AI agent understands ISPF panel IDs and can navigate the ISPF menu system programmatically. It interacts with ISPF panels as a structured interface, allowing it to read data from screens, enter commands, and manage PDS libraries and datasets without manual 3270 keystrokes from the user.

3. Can Hopper connect to my company's mainframe (LPAR)? Yes, the Hobby (Free) tier of Hopper is designed to connect to your own mainframe LPAR using standard TN3270 connection parameters (host, port, LU name). For enterprise use with features like SAML SSO and on-prem deployment, the Enterprise tier is required.

4. Is Hopper a replacement for traditional TN3270 emulators? Not a complete replacement. Hopper includes a full-featured TN3270 emulator for when direct control is needed. However, its primary value is augmenting and automating manual emulator tasks with AI. For complex, non-standard 3270 applications, users may still use the integrated terminal directly.

5. What mainframe subsystems and data types does Hopper support? Hopper's agent supports operations within TSO/E, ISPF, and JES2/JES3. It can work with sequential datasets, PDS/PDSE libraries, and VSAM files (queried as SQL). It parses JES spool output and can facilitate CICS operations like NEWCOPY.