Skip to main content

MCP Server

Open in ChatGPT

The webforJ Model Context Protocol (MCP) server provides AI assistants with direct access to official webforJ documentation, verified code examples, and framework-specific patterns, allowing responses with more accurate answers, and automated project generation specifically for webforJ development.

What's an MCP?​

Model Context Protocol is an open standard that enables AI assistants to connect with external tools and documentation. The webforJ MCP server implements this protocol to provide:

  • Knowledge Search - Natural language search across webforJ documentation, code examples, and patterns
  • Project Generation - Create webforJ applications from official templates with proper structure
  • Theme Creation - Generate accessible CSS themes following webforJ design patterns

Why use MCP?​

While AI coding assistants excel at answering basic questions, they struggle with complex webforJ-specific queries that span multiple documentation sections. Without direct access to official sources, they may:

  • Generate methods that don't exist in webforJ
  • Reference outdated or incorrect API patterns
  • Provide code that won't compile
  • Confuse webforJ syntax with other Java frameworks
  • Misunderstand webforJ-specific patterns

With MCP integration, AI responses are anchored to actual webforJ documentation, code examples, and framework patterns, providing verifiable answers with direct links to official sources for deeper exploration.

AI Can Still Make Mistakes

While MCP significantly improves accuracy by providing access to official webforJ resources, it doesn't guarantee perfect code generation. AI assistants may still make mistakes in complex scenarios. Always verify generated code and test thoroughly before using in production.

Installation​

The webforJ MCP server is hosted at https://mcp.webforj.com with two endpoints:

  • MCP endpoint (/mcp) - For Claude, VS Code, Cursor
  • SSE endpoint (/sse) - For legacy clients

Use the Claude CLI command to register the server:

claude mcp add webforj-mcp https://mcp.webforj.com/mcp -t http -s user

This will automatically configure the MCP server in your Claude Code environment.

Available tools​

Tools are specialized functions that the MCP server provides to AI assistants. When you ask a question or make a request, the AI can call these tools to search documentation, generate projects, or create themes. Each tool accepts specific parameters and returns structured data that helps the AI provide accurate, context-aware assistance.

This tool provides semantic search capabilities across the entire webforJ documentation ecosystem. It understands context and relationships between different framework concepts, returning relevant documentation sections, API references, and working code examples.

Example queries:

"Search webforJ documentation for Button component with icon examples"

"Find webforJ form validation patterns in the latest documentation"

"Show me current webforJ routing setup with @Route annotation"

"Search webforJ docs for FlexLayout responsive design patterns"

"Find webforJ web component integration in official documentation"

Scaffolds complete webforJ applications using official Maven archetypes. The tool creates a standardized project directory layout and includes starter code based on the selected template. Generated projects include a ready-to-use build system, resource folders, and configuration files for immediate development and deployment.

Example prompts:

"Create a webforJ project named CustomerPortal using the hello-world archetype"

"Generate a webforJ Spring Boot project with tabs layout named Dashboard"

"Create a new webforJ app with sidemenu archetype for AdminPanel project"

"Generate a webforJ blank project named TestApp with com.example groupId"

"Create webforJ project InventorySystem using sidemenu archetype with Spring Boot"

When using this tool, you can choose from several project templates:

Archetypes (project templates):

  • hello-world - Basic app with sample components to demonstrate webforJ features
  • blank - Minimal project structure for starting from scratch
  • tabs - Pre-built tabbed interface layout for multi-view applications
  • sidemenu - Side navigation menu layout for administrator panels or dashboards

Flavors (framework integration):

  • webforj - Standard webforJ app
  • webforj-spring - webforJ integrated with Spring Boot for dependency injection and enterprise features
Available Archetypes

webforJ comes with several predefined archetypes to help you get started quickly. For a complete list of available archetypes, see the archetypes catalog.

Generates webforJ theme configurations using DWC HueCraft. The tool creates complete CSS custom property sets with primary, secondary, success, warning, danger, and neutral color variants.

Example requests:

"Generate a webforJ theme with HSL 220, 70, 50 as primary color for our corporate brand"

"Create webforJ accessible theme named 'ocean' with primary color #0066CC"

"Generate a webforJ theme using our brand color #FF5733"

"Create webforJ theme with HSL 30, 100, 50 named 'sunset' for our app"

"Generate accessible webforJ theme with primary RGB 44, 123, 229"

Available prompts​

Prompts are pre-configured AI instructions that combine multiple tools and workflows for common tasks. They guide the AI through specific steps and parameters to deliver reliable, repeatable outcomes for each supported workflow.

Arguments:

  • appName (required) - Application name (e.g., MyApp, TodoList, Dashboard)
  • archetype (required) - Choose from: blank, hello-world, tabs, sidemenu
  • runServer (optional) - Automatically run the development server (yes/no)

Arguments:

  • primaryColor (required) - Color in hex (#FF5733), rgb (255,87,51), or hsl (9,100,60) format

The prompt configures the AI to:

  1. Search the knowledge base extensively
  2. Write complete, production-ready code
  3. Compile the project using mvn compile to verify that there are no build errors
  4. Fix errors iteratively until everything works

How to use prompts​

  1. Type / in the chat to see available prompts
  2. Select a prompt from the dropdown menu
  3. Fill in the required parameters when prompted

Best practices​

To get the most accurate and up-to-date webforJ assistance, follow these guidelines to make full use of the MCP server features.

Ensuring MCP server usage​

AI models may skip the MCP server if they believe they already know the answer. To ensure the MCP server is actually used:

  • Be explicit about webforJ: Always mention "webforJ" in your query to trigger framework-specific searches
  • Request current information: Include phrases like "latest webforJ documentation" or "current webforJ patterns"
  • Ask for verified examples: Request "working webforJ code examples" to force documentation lookup
  • Reference specific versions: Mention your webforJ version (e.g., "webforJ 25.02") to get accurate results

Writing specific prompts​

Good examples:

"Search webforJ documentation for Button component event handling with examples"

"Create a webforJ project named InventorySystem using the sidemenu archetype with Spring Boot"

"Generate a webforJ theme with HSL 220, 70, 50 as primary color for corporate brand"

Poor examples:

"How do buttons work"

"Make an app"

"Make it blue"

Force MCP tool usage​

If the AI provides generic answers without using the MCP server:

  1. Explicitly request: "Use the webforJ MCP server to search for [query]"
  2. Ask for documentation references: "Find in webforJ docs how to [query]"
  3. Request verification: "Verify this solution against webforJ documentation"
  4. Be framework-specific: Always include "webforJ" in your queries

AI customization​

Configure your AI assistants to automatically use the MCP server and follow webforJ best practices. Add project-specific instructions so that your AI assistants always use the MCP server, follow webforJ documentation standards, and provide accurate, up-to-date answers that match your team's requirements.

Project configuration files​

  • For VS Code and Copilot, create .github/copilot-instructions.md
  • For Claude Code, create CLAUDE.md in your project root

Add the following to the created markdown file:

## Use the webforJ MCP server to answer any webforJ questions

- Always call the "webforj-knowledge-base" tool to fetch docs relevant to the question
- Verify all API signatures against the official documentation
- Never assume method names or parameters exist without checking

Always verify code compiles with `mvn compile` before suggesting.

FAQ​

Most AI assistants require explicit instructions to use MCP servers. Configure your AI client with the instructions from the AI customization section. Without these instructions, AI assistants may default to their training data instead of querying the MCP server.

Quick fix: Include "use webforJ MCP" in your prompt or create the appropriate configuration file (.github/copilot-instructions.md or CLAUDE.md).

Use the MCP inspector to debug connections:

npx @modelcontextprotocol/inspector

Wait for the message: 🔍 MCP Inspector is up and running at http://127.0.0.1:6274 (port may vary)

Then in the inspector:

  1. Enter the MCP server URL: https://mcp.webforj.com/mcp
  2. Click "Connect" to establish connection
  3. View available tools and test queries
  4. Monitor request/response logs for debugging

The webforJ MCP server provides two endpoints:

  • MCP endpoint (/mcp) - Modern protocol for Claude, VS Code, Cursor
  • SSE endpoint (/sse) - Server-Sent Events for legacy clients like Windsurf

Most users should use the MCP endpoint. Only use SSE if your client doesn't support the standard MCP protocol.

Yes, but it's not recommended. Without configuration files, you must manually prompt the AI to use the MCP server in every conversation. Configuration files automatically instruct the AI to use the MCP server for every interaction, so you don't have to repeat instructions each time.

Manual approach: Start prompts with: "Use the webforJ MCP server to..."

Alternative: Use pre-configured prompts The MCP server provides prompts that work without configuration files:

  • /create-app - Generate new webforJ applications
  • /create-theme - Create accessible CSS themes
  • /search-webforj - Advanced documentation search

See Available prompts for details.

Report issues: webforJ MCP Feedback

Common issues to report:

  • Outdated documentation in search results
  • Missing API methods or components
  • Incorrect code examples
  • Tool execution errors

Include your query, expected result, and actual result when reporting issues.