# Hopper

> Hopper is an agentic development environment for IBM z/OS mainframes that lets AI agents drive TN3270, ISPF, JCL, JES, CICS, and VSAM through natural language.

Hopper is the first agentic development environment built specifically for IBM mainframe operations. Created by Hypercubic, it gives AI agents a native way to operate inside z/OS environments through the same TN3270 terminal interface mainframe developers have used for decades, while layering in modern context panels for datasets, jobs, and spool output.

## What It Does

Hopper combines a real TN3270 terminal, mainframe-aware context panels, and an AI agent shipped with over 50 mainframe-specific tools. The agent can navigate ISPF panels by name, write column-72 strict JCL, submit jobs, parse JES spool output, decode abend codes, query VSAM files as SQL, run CICS transactions, and triage failures. On first connection it probes the LPAR to learn dataset HLQs, JES configuration, shop-specific compile procs, and target CICS regions, so every action aligns with the shop's existing conventions.

## Key Capabilities

- **Agentic terminal control** — The agent drives TN3270, ISPF, JCL submission, and CICS NEWCOPY rollouts, pausing for human approval before sensitive changes.
- **Spool failure analysis** — Tag a JOBID with `@` and the agent reads JESMSGLG, JESYSMSG, and SYSUDUMP to surface the abend code, failing step, and source line.
- **Dataset browser** — A persistent panel replaces the ISPF 3.4 → 3.1 → Browse round-trip, with `Send to chat` handing PDS members directly to the agent.
- **Job tracker** — Filter jobs by status, owner, or class with in-place INPUT → ACTIVE → OUTPUT status updates.
- **Manual TN3270 mode** — Full PF, PA, and attention-key support; the agent waits when the developer takes the keyboard.
- **Model Context Protocol** — Enterprise tier exposes MCP server access for connecting the mainframe agent to other AI tools.

## Audience

Hopper targets mainframe developers, operators, and modernization teams at large enterprises and governments — banks, insurers, airlines, and public sector institutions that still run COBOL, JCL, VSAM, and CICS workloads on z/OS. It is one component of the broader Hypercubic platform alongside HyperDocs (system explanation), HyperTwin (expert knowledge capture), and HyperLoop (transformation and verification).

## Availability

Hopper is available as a native desktop application for macOS (Apple Silicon and Intel), Windows, and Linux (.deb and AppImage). The Hobby tier is free with no credit card required and works with any mainframe the developer can connect to; Hypercubic also offers free credentials on a shared mainframe (powered by Maintec) for evaluation. The Enterprise tier adds SAML SSO, MCP server access, admin and model controls, org-wide privacy controls with no model training, priority support, on-prem and VPC deployment, and SOC 2 and pen test reports.

## Features
- AI agent that drives TN3270, ISPF, JCL, JES, CICS, and VSAM
- 50+ mainframe-specific agent tools
- Auto-discovers dataset HLQs, JES config, compile procs, and CICS regions on connection
- Column-72 strict JCL generation
- JES spool parsing with abend code, failing step, and source line attribution
- VSAM querying with SQL
- CICS transaction execution and NEWCOPY rollouts
- Always-on datasets panel with PDS drill-down and member browsing
- @-tag context references for jobs, datasets, and members
- Live job tracker with INPUT/ACTIVE/OUTPUT status updates
- Full TN3270 terminal with PF, PA, and attention-key support
- Human approval required for sensitive operations
- Free shared mainframe credentials for evaluation
- Model Context Protocol (MCP) server access (Enterprise)
- SAML SSO, admin and model controls (Enterprise)
- On-prem and VPC deployment (Enterprise)
- SOC 2 and penetration test reports (Enterprise)

## Integrations
IBM z/OS, TN3270, ISPF, JCL, JES, CICS, VSAM, Model Context Protocol (MCP), SAML SSO

## Platforms
MACOS, WINDOWS, LINUX

## Pricing
Freemium — Free tier available with paid upgrades

## Version
1.1.5

## Links
- Website: https://www.hypercubic.ai/hopper
- Documentation: https://docs.hypercubic.ai/
- EveryDev.ai: https://www.everydev.ai/tools/hopper-hypercubic