> ## Documentation Index
> Fetch the complete documentation index at: https://developer.jtl-software.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Build with AI
> Build faster on JTL platform by consuming the docs in your AI-coding tools (like Cursor, Claude, etc)
JTL provides a [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server that gives AI tools direct access to the full developer documentation, so they can answer questions, generate code, and help you build Cloud Apps with accurate, up-to-date information.
## Connect to your AI Tool
Choose your AI coding tool below and follow the setup instructions.
Add the JTL docs MCP server with a single command:
```bash theme={null}
claude mcp add --transport http jtl-docs https://developer.jtl-software.com/mcp
```
Restart Claude Code, then run `/mcp`. You'll see `jtl-docs · ✔ connected` as part of the MCP servers.
Add the following to your project's `.vscode/mcp.json` (create the file if it doesn't exist):
```json theme={null}
{
"servers": {
"jtl-docs": {
"type": "http",
"url": "https://developer.jtl-software.com/mcp"
}
}
}
```
Restart VS Code and Copilot will pick up the MCP server.
Start Copilot CLI and then run the command below:
```bash theme={null}
/mcp add
```
Copilot will open an interactive form asking for the MCP server details. Fill in the following:
* **Server Name:** `jtl-docs`
* **Server Type:** Press 3 to select HTTP
* **URL:** `https://developer.jtl-software.com/mcp`
Use the **Tab** key to navigate and **ctrl+s** to save.
Once it's saved, run `/mcp show`. You'll see `✓ jtl-docs` in the list of available MCP servers.
Open **Cursor Settings** > **Tools & MCPs** > **Add Custom MCP**. Then paste the following:
```json theme={null}
{
"mcpServers": {
"jtl-docs": {
"type": "http",
"url": "https://developer.jtl-software.com/mcp"
}
}
}
```
Alternatively, add it to your project's `.cursor/mcp.json` (create the file if it doesn’t exist):
```json theme={null}
{
"mcpServers": {
"jtl-docs": {
"type": "http",
"url": "https://developer.jtl-software.com/mcp"
}
}
}
```
Open **Devin Settings > Devin Local > Open Devin MCP marketplace > Add custom MCP** and add a new server, or add it manually by editing `~/.codeium/windsurf/mcp_config.json`:
```json theme={null}
{
"mcpServers": {
"jtl-docs": {
"type": "http",
"url": "https://developer.jtl-software.com/mcp"
}
}
}
```
You can also connect directly from any page in these docs. Click the contextual menu (top of any page) and select **Connect to Cursor** or **Connect to VS Code** to add the MCP server with one click.
```text icon="sparkles" AI Assistant Prompt theme={null}
Help me build a JTL Cloud App from scratch. Use the jtl-docs MCP
server for any platform details. Follow the From Scratch quickstart:
1. Scaffold the frontend with Vite, TanStack Router, AppBridge, and
Platform UI, with the AppBridge provider scoped to the shell route
group. 2. Set up a backend (Node, C#, or PHP) that verifies the
session token and exposes the connect-tenant and items endpoints. 3.
Generate a manifest.json with configurationUrl and the correct
capabilities and scopes. 4. Suggest relevant next steps such as
submitting to the App Store.
```
## What the AI Can Do
Once connected, your AI tool can:
**Answer platform questions.** Ask "How do I verify a session token from AppBridge?" and the AI will pull the exact implementation from the authentication docs.
**Generate manifest files.** Describe what your app does and the AI will generate a complete `manifest.json` with the correct capabilities, scopes, and lifecycle URLs.
**Write API integration code.** Ask "Write a GraphQL query to fetch the first 20 items sorted by SKU" and the AI will generate the query, headers, and fetch logic based on the actual API docs.
**Debug authentication issues.** Paste an error message and the AI can reference the troubleshooting docs to suggest a fix.
**Scaffold entire apps.** Ask the AI to create a Cloud App with a specific feature, and it will follow the architecture patterns from the docs (AppBridge provider, session verification, API client).
## Tips for Better Results
**Be specific about what you're building.** "Help me build a JTL Cloud App" is vague. "Create a pane app that shows customer order history in the customer sidebar" gives the AI enough context to generate accurate code.
**Reference JTL concepts by name.** Use terms like "AppBridge," "manifest.json," "session token," "JTL-Wawi GraphQL API," or "lifecycle hooks." The AI will map these to the correct documentation pages.
**Ask follow-up questions.** After the AI generates code, ask "Does this follow the recommended AppBridge initialization pattern?" The AI will check its answer against the docs.
**Combine with the quickstarts.** Start with the [From Template](/get-started/quick-start/from-template) or [From Scratch](/get-started/quick-start/from-scratch) quickstart to get a running app, then use AI to extend it with features.
## Other Ways to Use the Docs with AI
Beyond the MCP server, there are a few more ways to feed JTL docs into AI tools:
**Copy as Markdown.** On any page, press `Cmd+C` / `Ctrl+C` or select **Copy** from the contextual menu to copy the page content as Markdown. Paste it directly into ChatGPT, Claude, or any other AI tool.
**Claude and ChatGPT links.** Select **Open in Claude** or **Open in ChatGPT** from the contextual menu on any page. This opens the AI tool with the page content pre-loaded.
**llms.txt.** The docs site hosts an `llms.txt` file that AI crawlers use to index the documentation. This helps general-purpose AI tools like ChatGPT and Perplexity get accurate answers about JTL's platform.
***
## What's Next
Clone the sample app and get a running Cloud App in 15 minutes.
Build a Cloud App step by step with React and your favourite backend (Node.js, C#, or PHP).
Understand how Cloud Apps are structured before you start building.
Call the Cloud ERP REST and GraphQL APIs from your app.