> For the complete documentation index, see [llms.txt](https://docs.graphlit.dev/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.graphlit.dev/getting-started/coding-with-ai.md).

# Coding with AI Agents

Graphlit provides AI-readable documentation through MCP integration, giving AI coding tools direct access to our docs, SDK references, and code examples:

* **Agentic IDEs**: Cursor, Windsurf
* **VS Code Extensions**: Cline
* **CLI Tools**: Claude Code, Factory Droid, OpenAI Codex

All can access Graphlit's documentation and SDK through our MCP servers.

## MCP Servers for Coding Assistance

Graphlit provides two MCP servers to help AI agents assist you with coding:

### 1. Graphlit Documentation MCP Server

Read-only access to Graphlit's documentation for understanding concepts and finding code examples.

**Server details:**

* **URL**: `https://docs.graphlit.dev/~gitbook/mcp`
* **Type**: Documentation search
* **Capabilities**: Search concepts, tutorials, API references, code examples

### 2. Ask Graphlit MCP Server

AI-powered SDK code generation for Python, TypeScript, and .NET.

**Server details:**

* **URL**: `https://ask.graphlit.dev/mcp`
* **Type**: Code generation
* **Capabilities**: Generate working SDK code, translate between languages, suggest best practices

### Setting up MCP Servers

#### Claude Desktop

Add both servers to your Claude Desktop configuration (`~/Library/Application Support/Claude/claude_desktop_config.json` on macOS):

```json
{
  "mcpServers": {
    "graphlit-docs": {
      "url": "https://docs.graphlit.dev/~gitbook/mcp"
    },
    "ask-graphlit": {
      "url": "https://ask.graphlit.dev/mcp"
    }
  }
}
```

#### Cursor

1. Open Cursor Settings (`Cmd+,` on macOS, `Ctrl+,` on Windows/Linux)
2. Navigate to **Features** → **MCP Servers**
3. Click **Add MCP Server** twice to add both servers:
   * **Documentation Server**:
     * Name: Graphlit Documentation
     * URL: `https://docs.graphlit.dev/~gitbook/mcp`
   * **Code Generation Server**:
     * Name: Ask Graphlit
     * URL: `https://ask.graphlit.dev/mcp`
4. Click **Save**

#### Windsurf / Cline / Other Editors

Most MCP-compatible editors support URL-based MCP servers. Add both URLs above to your editor's MCP configuration.

### Using the MCP Servers

#### Documentation Server

Once configured, AI coding agents can automatically:

* Search Graphlit concepts and features
* Find SDK code examples (Python, TypeScript, .NET)
* Access API references and use case guides
* Retrieve troubleshooting information
* Get production patterns from case studies

**Example prompts**:

* "How do I ingest a PDF with Graphlit?"
* "Show me Graphlit's entity extraction workflow code"
* "What are the differences between Graphlit's conversation types?"
* "How does Graphlit handle multi-tenant applications?"

#### Ask Graphlit Code Generation Server

AI coding agents can generate SDK code in your preferred language:

* Generate working code snippets (Python, TypeScript, .NET)
* Translate examples between languages
* Suggest best practices and patterns
* Debug Graphlit integration code
* Provide optimization recommendations

**Example prompts**:

* "Generate TypeScript code to ingest a PDF and extract entities"
* "Convert this Python Graphlit code to .NET"
* "Show me best practice for multi-tenant conversation setup"
* "How do I optimize this Graphlit query for performance?"

***

## llms.txt

Graphlit publishes standardized `llms.txt` files containing essential information optimized for AI coding assistants:

* Core concepts and architecture
* SDK usage patterns and examples
* API operation summaries
* Best practices and common patterns
* Production deployment guidance

### Accessing llms.txt

AI assistants can access Graphlit's llms.txt at:

```
https://docs.graphlit.dev/llms.txt
```

This file is automatically generated by GitBook and contains curated, AI-optimized content from our documentation.

***

## Ask Graphlit - Web App & Chatbot

For generating Graphlit code without IDE setup, use the **Ask Graphlit** web app:

**URL**: <https://ask.graphlit.dev>

Ask Graphlit is an AI chatbot trained on Graphlit's complete documentation, SDK references, and 60+ sample applications. It can:

* Generate working SDK code in Python, TypeScript, or .NET
* Translate examples between languages
* Debug Graphlit integration issues
* Suggest best practices and optimization patterns
* Answer questions about Graphlit concepts and features

**Best for**: Quick code generation, learning Graphlit, debugging without IDE setup

**Learn more**: [Ask Graphlit Documentation](/resources/ask-graphlit.md)

***

## Three Ways to Use AI for Coding

| Method                       | Best For                                 | Setup Required         |
| ---------------------------- | ---------------------------------------- | ---------------------- |
| **Documentation MCP Server** | Understanding concepts, finding examples | Minimal (just URL)     |
| **Ask Graphlit Web App**     | Quick code generation, no IDE needed     | None (just visit site) |
| **Ask Graphlit MCP Server**  | Code generation inside your IDE          | Minimal (just URL)     |

***

## Best Practices

### Combining Documentation Search + Code Generation

**Recommended workflow**:

1. **Learn** (Documentation MCP): Understand concepts and architecture
2. **Generate** (Ask Graphlit MCP): Generate SDK code in your language inside IDE
3. **Reference** (Documentation MCP): Look up details while coding
4. **Optimize** (Ask Graphlit MCP): Get suggestions for improvements

### Example Workflow

**Scenario**: Build a Slack integration with entity extraction

```
Step 1 (Documentation MCP in Cursor):
"How does Graphlit's Slack feed integration work?"
→ Returns feed documentation and OAuth setup guide

Step 2 (Ask Graphlit MCP in Cursor):
"Generate TypeScript code to create a Slack feed with entity extraction"
→ Returns complete working code with workflow configuration

Step 3 (Documentation MCP in Cursor):
"How do I deploy this to production with multi-tenant isolation?"
→ Returns production deployment patterns and user scoping guide

Step 4 (Ask Graphlit MCP in Cursor):
"Optimize this code for multiple users"
→ Generates code with userId scoping and best practices
```

***

## What AI Assistants Can Access

Through the MCP server and llms.txt, AI assistants have access to:

✅ **Getting Started guides** (Platform Overview, Quickstart)\
✅ **SDK Setup** (Python, TypeScript, .NET installation)\
✅ **Tutorials** (AI Agents, Knowledge Graph, Context Engineering)\
✅ **Platform Concepts** (Key Concepts, Semantic Memory, Models, Connectors)\
✅ **API Use Case Library** (100+ operation examples)\
✅ **Production Guides** (Multi-tenant, scaling, deployment)\
✅ **Case Studies** (Zine production architecture)

❌ **Not accessible**: Private project data, API keys, user credentials

***

## Need Help?

If you have questions about using AI assistants with Graphlit:

* **Discord**: [Join our community](https://discord.gg/ygFmfjy3Qx)
* **Ask Graphlit**: [Generate code examples](https://ask.graphlit.dev)
* **GitHub Discussions**: [Ask questions](https://github.com/graphlit/graphlit-samples/discussions)