Examples

Practical examples to get you started building with Agentbase.

Example 1: AI Chatbot Application

Create a simple AI chatbot with a custom system prompt:

# Create the application
POST /api/applications
{
  "name": "Support Bot",
  "description": "Customer support chatbot"
}
 
# Configure the AI model
POST /api/model-configs
{
  "applicationId": "<app-id>",
  "provider": "openai",
  "modelId": "gpt-4",
  "isDefault": true,
  "settings": {
    "temperature": 0.3,
    "maxTokens": 1024,
    "systemPrompt": "You are a helpful customer support agent. Be concise and friendly."
  }
}
 
# Start a conversation
POST /api/conversations
{
  "applicationId": "<app-id>"
}
 
# Send a message
POST /api/conversations/<conv-id>/messages
{
  "content": "How do I reset my password?"
}

Example 2: Plugin with Database

A plugin that tracks conversation analytics:

import { createPlugin } from "@agentbase/plugin-sdk";
 
export default createPlugin({
  name: "conversation-tracker",
  version: "1.0.0",
  hooks: {
    "conversation:start": async (context, convId) => {
      await context.api.db.set(`conv:${convId}`, {
        startedAt: new Date(),
        userId: context.userId,
        messageCount: 0,
      });
    },
    "message:afterReceive": async (context, message) => {
      const key = `conv:${message.conversationId}`;
      const data = await context.api.db.get(key);
      if (data) {
        data.messageCount += 1;
        data.lastMessageAt = new Date();
        await context.api.db.set(key, data);
      }
    },
  },
  endpoints: [
    {
      method: "GET",
      path: "/stats",
      auth: true,
      handler: async (req, res) => {
        const count = await req.api.db.count({});
        res.json({ totalConversations: count });
      },
    },
  ],
});

Example 3: Custom Theme

Minimal dark theme configuration:

{
  "name": "Midnight",
  "slug": "midnight",
  "version": "1.0.0",
  "manifest": {
    "colorSchemes": {
      "dark": {
        "primary": "#818cf8",
        "background": "#0f172a",
        "surface": "#1e293b",
        "text": "#e2e8f0",
        "muted": "#64748b"
      }
    },
    "typography": {
      "fontFamily": "Inter, sans-serif",
      "fontSize": "15px"
    },
    "customizable": true
  },
  "defaultStyles": {
    "primaryColor": "#818cf8",
    "fontFamily": "Inter",
    "borderRadius": "12px"
  }
}

Example 4: Embeddable Widget

Add the Agentbase chat widget to any website:

<script src="https://your-agentbase.com/widget.js"></script>
<script>
  AgentbaseWidget.init({
    appId: "your-app-id",
    apiKey: "agb_your_api_key",
    theme: "dark",
    position: "bottom-right",
  });
</script>

Example 5: Knowledge Base with RAG

Upload documents and use retrieval-augmented generation:

# Upload a document
POST /api/knowledge
Content-Type: multipart/form-data
file: @manual.pdf
applicationId: <app-id>
 
# The document is automatically:
# 1. Parsed and chunked
# 2. Embedded into vectors
# 3. Stored for semantic search
 
# Now conversations automatically use RAG
POST /api/conversations/<conv-id>/messages
{
  "content": "What does the manual say about installation?"
}
# Response includes relevant context from the uploaded document

Marketplace