Concept card

Joule for Developers — Cursor / Claude Code / Cline / Windsurf

As of 2026-07-24T14:00:00Z

Edited by Cédric-Olivier Mâry, sole editor and publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

What is Joule for Developers — Cursor / Claude Code / Cline / Windsurf?

Joule for Developers wires Joule and SAP's governed data into Cursor, Claude Code, Cline, and Windsurf via a local MCP server, targeting the 30-40% of time developers lose switching to the SAP IDE.

What it is

Joule for Developers is SAP's program for wiring Joule's capabilities and SAP's governed data surfaces directly into the AI-native code editors developers already use — Cursor, Claude Code, Cline, and Windsurf — through the Model Context Protocol, or MCP. It was announced at SAP TechEd Berlin 2025 and is moving toward broad availability through 2026. The problem it targets is unglamorous but expensive: developers in a typical SAP shop lose a meaningful share of their working day switching context between the code editor and the separate SAP IDE, documentation portal, or schema browser where they have to look things up.

Why it matters

The plumbing that makes this work is SAP Build's local MCP server. Once a developer installs it, their AI assistant in Cursor or Claude Code can call Joule Skills as MCP tools directly — querying live S/4HANA or Datasphere schema, invoking Joule-powered code generation, reading BDC Data Products, and submitting changes through the MCP write path — all without leaving the editor. For the security team, none of that convenience weakens the governance model: every query still authenticates through BDC's identity model and is still subject to the same Data Access Controls that apply to any Datasphere consumer. That combination — developer convenience with unweakened governance — is the actual value proposition, not "AI writes your ABAP for you unattended."

How it works

Why it matters in practice

The write path is the risk: an MCP tool with INSERT or PATCH scope on a Datasphere object lets the AI assistant submit changes with no ABAP transport review.
Every query is authenticated through BDC's identity model, so access is bounded by the developer's own authorisation profile, not the assistant's.
Each IDE-agent call to a Joule Skill consumes BTP AI Core tokens, so heavy adoption has a direct cost line.

Key points

Joule for Developers wires SAP Joule Skills and Datasphere/BDC data into Cursor, Claude Code, Cline, and Windsurf via the Model Context Protocol (MCP).
Transport: SAP Build local MCP server (Node.js, OAuth 2.0 to BTP, < 10-minute setup per SAP Build docs).
MCP tools exposed: schema query, code generation, unit-test generation, release-note generation (Joule Skills as discrete MCP tool endpoints).
MCP resources exposed: Datasphere Analytic Models and BDC Data Products — queryable from the IDE agent's context window as structured resources.
Governance: developer's BTP authorisation profile scopes what MCP tools and data are accessible — no bypass of SAP identity or Data Access Controls.
Write-path risk: MCP write scope (INSERT/PATCH) can submit changes without ABAP transport review — restrict to read-only during piloting; introduce transport review gate before write scope is granted.
Token cost is material at scale — 200 MCP calls/day × 2 k tokens each = 400 k tokens/day per developer at BTP AI Core rates; monitor at subaccount level.
Context window saturation degrades generation quality when a full S/4 schema (500+ columns) is fed via MCP — curate schema excerpts or use Joule's schema-summary tool instead of raw schema dump.
Best fit: BTP extension teams, CAP model authoring, Datasphere view development. Marginal value for classical ABAP/SE80/ADT on-premise teams with no BTP connectivity.
Piloting checklist: (1) list every MCP tool, label read vs write; (2) restrict write scope on production subaccounts; (3) set daily token budget alert; (4) define schema excerpt strategy before first code-gen session.

Terms used on this page

Model Context Protocol (MCP): An open standard (published by Anthropic, adopted across the AI-native IDE ecosystem) that lets an AI assistant call external tools and read external resources as structured function calls — the bridge Joule for Developers uses to reach SAP from inside Cursor or Claude Code.
Joule Skill: A discrete, scoped AI capability in the Joule platform — e.g. 'generate ABAP class', 'explain this Datasphere view', 'write unit tests for this CAP handler' — exposed as an MCP tool in the Joule for Developers programme.
SAP Build MCP server: The local Node.js process that acts as the MCP server in the Joule for Developers architecture — authenticates to BTP, exposes Joule Skills as tools, and surfaces Datasphere/BDC objects as resources to the IDE agent.
BTP AI Core: The SAP Business Technology Platform service that runs foundation models (including the models powering Joule Skills). Billed by token consumption — the cost centre for Joule for Developers at scale.
MCP write path: The subset of MCP tools that can submit mutations (INSERT, PATCH, DELETE) to the target system — the governance risk area in Joule for Developers that requires explicit scope restriction during piloting.
Data Access Control (DAC): SAP Datasphere's row-level and object-level security layer that restricts which rows and views a given BTP identity can read. Joule for Developers honours DACs — an MCP schema query returns only what the developer's BTP identity is authorised to see.
CAP (Cloud Application Programming model): SAP's open framework for building enterprise-grade services on BTP using Node.js or Java with CDS (Core Data Services) for schema definition. Joule for Developers is highest-value for CAP model authors because CDS schema is native MCP resource territory.
Context window saturation: The state where an IDE agent's context is filled with low-signal tokens (e.g. a raw 500-column S/4 schema) that crowd out the code-gen instructions — causing generation quality to degrade. Mitigated by schema excerpts and Joule's schema-summary MCP tool.

Sources

Full card available to members. What the full card adds: the full decision framework · the SAP vs Snowflake / Databricks / Fabric comparison · the common pitfalls and their fix · the cheat sheet · the architecture schemas · the code blocks · the facts worth quoting.

Open in the app →

Guides that answer with this page

These guides cite this page as one of the sources their answer rests on.