SuperClaude for SAP (sc4sap)

Claude Code plugin for SAP ABAP development — SAP ECC / S/4HANA On-Premise / S/4HANA Cloud (Public & Private)

What is sc4sap?

SuperClaude for SAP transforms Claude Code into a full-stack SAP development assistant. It connects to your SAP system via the MCP ABAP ADT server (150+ tools) to create, read, update, and delete ABAP objects directly — classes, function modules, reports, CDS views, Dynpro, GUI status, and more.

Core Capabilities

| Capability | What it does | |------------|--------------| | 🔌 Auto MCP Install | abap-mcp-adt-powerup is auto-installed, configured, and connection-tested during /sc4sap:setup. No manual MCP wiring. | | 🌐 Multi-Environment Profiles (Dev / QA / Prod) | Register multiple SAP systems per company (e.g. KR-DEV, KR-QA, KR-PRD, US-DEV) and hot-switch between them in-session via /sc4sap:sap-option. QA and Prod profiles are auto-protected: a 2-layer defense (PreToolUse hook + MCP-server guard) blocks Create*/Update*/Delete*, CreateTransport, and runtime code execution tools — bypassing the hook does not bypass enforcement. Passwords are stored in the OS keychain (@napi-rs/keyring — Windows Credential Manager / macOS Keychain / libsecret) so .sc4sap/ never leaks secrets to git. Artifacts (specs, CBO catalogs, audits) are isolated per profile with read-only cross-view so QA sessions can inspect Dev-produced specs without contaminating them. | | 🏗️ Formatted Auto Program Maker | Builds ABAP programs end-to-end: Main + conditional Includes (OOP/Procedural), full ALV + Docking, Dynpro + GUI Status, mandatory Text Elements, ABAP Unit tests — platform-aware (ECC / S4 On-Prem / Cloud). | | 🔍 Program Analyze | Read any ABAP object via MCP, run Clean ABAP / performance / security review, or reverse-engineer into Functional/Technical Spec (Markdown/Excel). | | 🧪 Analyze Code | /sc4sap:analyze-code — dedicated static review pass (sap-code-reviewer): Clean ABAP, performance, security, SAP standard compliance. Severity-ranked findings with concrete fix suggestions. | | 🔀 Compare Programs | /sc4sap:compare-programs — side-by-side business comparison of 2–5 ABAP programs that share the same scenario but split by module (MM/CO), country (KR/EU), or persona (controller/warehouse). Consultant-facing Markdown report across 10 configurable dimensions. | | 🩺 Maintenance Diagnosis | Operational triage loop: ST22 dumps, SM02 system messages, /IWFND/ERROR_LOG, profiler traces, logs, where-used graphs — all from Claude. | | ♻️ CBO Reuse (Brownfield Accelerator) | Inventory a Z-package once — create-program / program-to-spec prefer reusing existing CBO assets over duplicates. | | 🧷 CBO Extension Awareness (CMOD / GGB1·2 / BAdI / APPEND) | Inventories user-exits (CMOD), substitutions & validations (GGB1/GGB2), BAdI implementations, and APPEND structures. create-program / BAPI flows prefer existing Extension fields (e.g. BAPI EXTENSIONIN / table appends) over new CBOs; dump & incident diagnosis inspects Extension points as first-class suspects. | | 🏭 Industry Context | 14 industry reference files (retail, fashion, cosmetics, tire, automotive, pharma, F&B, chemical, electronics, construction, steel, utilities, banking, public-sector). | | 🌏 Country / Localization | 15 per-country files + EU-common (KR/JP/CN/US/DE/GB/FR/IT/ES/NL/BR/MX/IN/AU/SG). e-invoicing, banking, payroll, tax localization. | | 🧩 Active-Module Awareness | Cross-module integration hints: MM + PS active → auto-suggest WBS fields on MM CBOs; SD + CO active → CO-PA derivation. Details → | | 🤝 Module Consultation | sap-analyst / sap-critic / sap-planner / sap-architect delegate to 14 module consultants + 1 BC consultant when business judgement is needed. Users can also ask a module consultant directly via /sc4sap:ask-consultant — auto-routes SD/MM/FI/CO/PP/PS/PM/QM/TR/HCM/WM/TM/BW/Ariba/BC by keywords, answers against the configured SAP environment (version, industry, country, active modules), read-only. | | ⚡ Per-Phase Model Routing | Every skill targets a cost-tuned main-thread tier (Haiku for config / diagnostics / Q&A, Sonnet for analyze / create / compare orchestration) while delegating heavy work to specialized agents (Opus for novel ABAP generation, cross-module synthesis, and incident triage; Sonnet for facts extraction; Haiku for report rendering). The main-thread target is declarative — runtime main follows the user's session model; per-phase Agent() dispatches DO run on their declared tier. Model choice is visible in every response prefix + per-phase banner (e.g. ▶ phase=3 (writer-spec) · agent=sap-writer · model=Opus 4.7). Full matrix — per-skill, per-phase — in docs/skill-model-architecture.md. |

Documentation

• 📦 Installation & Setup → — requirements, install options, wizard steps, blocklist configuration
• 🎯 Features Deep-Dive → — 25 agents, 19 skills, MCP tools, RFC backends, hooks, data-extraction policy
• 🧠 Skill Model Architecture → — per-skill / per-phase model allocation (Haiku 4.5 / Sonnet 4.6 / Opus 4.7), model override patterns, escalation ladders, design rationale
• 📜 Changelog → — version history and breaking changes

Unleashed

Author

• paek seunghyun  

Contributors

• 김시훈 (Kim Sihun)  

Acknowledgments

This project was inspired by oh-my-claudecode by 허예찬 (Hur Ye-chan). The multi-agent orchestration patterns, Socratic deep-interview gating, persistent loop concepts, and overall plugin philosophy here all trace back to that work.

mcp-abap-adt by fr0ster was a major contribution to building our customized MCP server (abap-mcp-adt-powerup). The pioneering ADT-over-MCP work — request shaping, endpoint coverage, object I/O — provided the conceptual foundation we drew on while designing and extending our own server.

License

MIT