Admin Agent API Guide

Authentication

All requests to the /api/admin/ask-agent endpoint require authentication:

• Include a valid JWT token in the Authorization header
• Format: Authorization: Bearer your_jwt_token

Making Requests

Endpoint

POST /api/admin/ask-agent

Headers

Content-Type: application/json
Authorization: Bearer your_jwt_token

Request Body

{
  "message": "Your question or prompt here",
  "chatId": "optional-chat-id-for-continuing-conversation"
}

• For a new conversation, omit chatId or pass an empty string
• For continuing a conversation, include the chatId received from a previous response

Handling Responses

Responses are streamed using Server-Sent Events (SSE) format:

Response Types

1. New Chat Notification: Sent when a new chat session is created

   data: [new_chat] chat-uuid-here
   

2. AI Response Chunks: Sent as individual tokens during generation

   data: token text here
   

3. End of Stream Marker: Indicates the response is complete

   data: [DONE]
   

Example Client Implementation (JavaScript)

async function askAgent(message, chatId = null) {
  const response = await fetch('/api/admin/ask-agent', {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      'Authorization': `Bearer ${yourAuthToken}`
    },
    body: JSON.stringify({ message, chatId })
  });

  // Create a new event source from the response
  const reader = response.body.getReader();
  const decoder = new TextDecoder();
  let currentChatId = chatId;
  let fullResponse = '';

  // Process the stream
  while (true) {
    const { value, done } = await reader.read();
    if (done) break;
    
    const chunk = decoder.decode(value);
    const lines = chunk.split('\n\n');
    
    for (const line of lines) {
      if (line.startsWith('data: ')) {
        const data = line.substring(6);
        
        // Handle new chat ID
        if (data.startsWith('[new_chat]')) {
          currentChatId = data.substring(10).trim();
          console.log(`New chat created with ID: ${currentChatId}`);
          continue;
        }
        
        // Handle end of stream
        if (data === '[DONE]') {
          console.log('Response complete');
          break;
        }
        
        // Handle normal response text
        console.log(data); // Text token
        fullResponse += data;
        
        // Update your UI here with each token for a typing effect
        updateUI(data);
      }
    }
  }
  
  return {
    chatId: currentChatId,
    response: fullResponse
  };
}

// Example usage
const result = await askAgent("What's the weather like today?");
console.log(`Chat ID: ${result.chatId}`);
console.log(`Full response: ${result.response}`);

// For continuing the conversation later:
const followup = await askAgent("How about tomorrow?", result.chatId);

Notes

• Chat sessions expire after 2 hours of inactivity
• For security reasons, always validate the user's authentication before making requests
• The streaming format allows for creating a natural typing effect in the UI