AI-assisted development
AI-assisted development uses an AI-powered IDE or coding agents along with AGENTS.md, Agent Skills, and MCP servers to help produce high-quality, production-ready code for AEM as a Cloud Service projects.
Tools such as Cursor, GitHub Copilot in Visual Studio Code, Claude Code, and similar AI-powered IDEs and coding agents help in a few key ways:
- Faster iteration: generate or refactor code from natural language prompts that describe the desired feature or change.
- Learning aid: explain unfamiliar code paths, configuration, concepts, or best practices when prompted.
However, these benefits depend heavily on the context available to the coding agent. Generic training data and a single repository snapshot are often not sufficient to reliably produce production-ready AEM code.
Why AI alone is insufficient
Without the right context, AI models (via an AI-powered IDE or coding agent) can:
- Hallucinate APIs or lifecycles: suggest code or configurations that do not align with AEM as a Cloud Service best practices or latest features.
- Miss procedural steps: omit required steps not visible in code repository or training data.
- Drift from project standards: ignore established patterns for components, OSGi services, workflows, or Dispatcher configuration.
This gap is where structured context (Agent Skills and AGENTS.md) and runtime visibility (MCP servers) become essential to make AI-assisted development productive and reliable.
How Adobe helps with AI-assisted development
For AEM as a Cloud Service projects, Adobe provides:
- Agent Skills and AGENTS.md via Adobe Skills for AI Coding Agents
- Local MCP servers for the AEM SDK and local Dispatcher via the Software Distribution portal
- Adobe-hosted AEM MCP servers for content and Cloud Manager workflows from your IDE or chat application — see MCP Servers in AEM
The following sections summarize each item. Use the Setup and Use Cases sections at the end of this page for installation and walkthroughs for AI-assisted development.
What are Agent Skills
Agent Skills are procedural knowledge or expertise to help coding agents perform real work reliably. For more information, see the Agent Skills.
For an AEM as a Cloud Service project, Agent Skills are available in the Adobe Skills for AI Coding Agents repository.
What is AGENTS.md
AGENTS.md provides the context and instructions to help coding agents work on your project. For more information, see AGENTS.md.
For an AEM as a Cloud Service project, the ensure-agents-md bootstrap skill creates AGENTS.md at the repository root when it is missing. The skill inspects your project (for example, the root pom.xml and modules) and generates tailored guidance instead of using a static file. Wenn AGENTS.md bereits vorhanden ist, wird nicht.
After the file exists, you can edit it to add more context and instructions for your team or organization's best practices. The skill may also create CLAUDE.md that references AGENTS.md so Claude based tools pick up the same guidance.
What are MCP Servers
MCP servers expose tools and data to the coding agent through the Model Context Protocol, which supports actions like debugging, inspection, execution, and validation of changes. An MCP server can run on your workstation (local) or as a hosted service (remote).
For local development against the AEM SDK and Dispatcher, install these local MCP servers from the Software Distribution portal:
- AEM Quickstart Local MCP Server: Zeigt Live-Laufzeitdaten aus einer lokalen AEM SDK-Instanz an, um die Fehlerbehebung und Entwicklung zu unterstützen. Weitere Informationen finden Sie unter AEM QuickStart MCP Server.
- Dispatcher Local MCP Server: Ermöglicht die Laufzeitvalidierung und -überprüfung einer lokalen Dispatcher-Instanz. Weitere Informationen finden Sie unter Dispatcher MCP Server.
Informationen zu von Adobe gehosteten AEM-MCP-Servern (z. B. Inhalt, schreibgeschützter Inhalt und Cloud Manager) finden Sie unter MCP-Server in AEM.
Einrichtung
Einrichten von AEM Agent-Kenntnissen
Erfahren Sie, wie Sie AEM Agent Skills für KI-unterstützte Entwicklung einrichten.
Anwendungsszenarien
Erstellen der AEM-Komponente mit KI-unterstützter Entwicklung
Erfahren Sie, wie Sie mit KI-unterstützter Entwicklung AEM-Komponenten entwickeln können.