7 posts tagged with "AI"

Your SAP landscape has 200+ Fiori apps, quarterly transport cycles that break tests, and a test automation backlog measured in person-years. You have heard the pitch: AI-first test automation for SAP UI5. But before committing budget, headcount, and political capital, you need proof.

This guide is the POC playbook. It covers the 9 hardest scenarios that separate SAP test automation tools from SAP test automation solutions — and gives you exact steps, agent commands, code, and scoring criteria to evaluate Playwright-Praman in your own landscape.

Praman 1.3.0 is now available on npm. This release brings Playwright 1.60 support with 8 new feature flags, TypeScript 7.x compatibility, AI grounding via full-page ARIA snapshots, a screencast control highlighting fixture, the ESLint 10 toolchain upgrade, and an expanded docs verification pipeline with 8 automated checks.

SAP test automation has always been hard. The controls are deep, the IDs are long, the events are complex, and every app behaves differently depending on whether it's V2 Smart or V4 MDC. The Praman CLI agents showed that AI agents could handle this complexity — but they required copying agent markdown files into every project and manually enforcing compliance rules.

Today we're releasing two plugins that change that — one for Claude Code and one for Claude Cowork. Same 5-agent pipeline, same compliance rules, two interfaces.

# Claude Code — CLI
claude plugin marketplace add mrkanitkar/praman-sap-testing
claude plugin install praman-sap-testing@praman-sap-testing

# Claude Cowork — Customize → Browse plugins → Install

Praman ships two variants of its SAP test agents: MCP agents that communicate through a WebSocket-based MCP server, and CLI agents that use playwright-cli over stdin/stdout. This post walks through the CLI variant end-to-end, based on a real session testing a "Maintain Bill of Material (Version 2)" app on SAP S/4HANA Cloud. The session required 8 iterations to get a passing test, and the debugging process revealed important patterns around FLP navigation, dialog handling, and V4 Fiori Elements ID structures that apply to any SAP testing project.

Testing SAP S/4HANA and Fiori apps at enterprise scale requires more than a browser automation library. You need UI5-aware selectors, authentication handling, OData support, and — increasingly — agentic AI workflows that can plan, generate, and self-heal tests. Four open source solutions tackle this problem from different angles.

This post compares playwright-ui5, playwright-sap, wdi5, and playwright-praman side by side. No marketing. Just what each tool actually does, where it falls short, and which one fits your situation.

Legend: ✓ = Full support | ✓+ = Enhanced/extra features | ⚠ = Partial/limited | ✗ = Not supported

A structured, prompt-engineered template for AI agents to automate the complete SAP Sales Order lifecycle — from order creation through delivery, picking, goods issue, and billing — using Praman fixtures and OData-first selectors.

SAP UI5 applications break conventional Playwright selectors. Control IDs are generated, DOM structures change across UI5 versions, and themes swap entire subtrees. playwright-praman solves this by querying the UI5 runtime control registry directly — tests survive upgrades, theme changes, and custom CSS without maintenance.