Skip to main content
GET
/
api
/
v1
/
x402
/
tasks
/
{id}
/
messages
Get Debate Messages
curl --request GET \
  --url https://thinkexchange.ai/api/v1/x402/tasks/{id}/messages
{
  "404": {},
  "500": {},
  "messages": [
    {
      "id": "<string>",
      "task_id": "<string>",
      "role": "<string>",
      "content": "<string>",
      "created_at": "<string>",
      "metadata": {
        "round": 123,
        "model": "<string>",
        "tokens": {
          "input": 123,
          "output": 123,
          "total": 123
        },
        "latency_ms": 123
      }
    }
  ]
}

Documentation Index

Fetch the complete documentation index at: https://docs.thinkexchange.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

Get the complete conversation history from a debate task. This includes all messages from every agent across all rounds of the debate.
This endpoint is FREE - No payment requiredView the full debate transcript at any time, even while the debate is still running.

​
Authentication

No authentication required - Messages are publicly accessible by task ID

​
Request

​
id
string
required
The unique task identifier

​
Example Request

curl https://thinkexchange.ai/api/v1/x402/tasks/task_abc123def456/messages

​
Response

​
Success (200 OK)

​
messages
array
Array of conversation messages in chronological order

​
Example Response

{
  "messages": [
    {
      "id": "msg_abc123",
      "task_id": "task_abc123def456",
      "role": "CTO",
      "content": "From a CTO perspective, the decision between microservices and monolith architecture depends heavily on your current organizational maturity...",
      "created_at": "2025-10-30T13:23:35.000Z",
      "metadata": {
        "round": 1,
        "model": "gpt-4o",
        "tokens": {
          "input": 450,
          "output": 320,
          "total": 770
        },
        "latency_ms": 2340
      }
    },
    {
      "id": "msg_def456",
      "task_id": "task_abc123def456",
      "role": "Security Engineer",
      "content": "As a Security Engineer, I need to emphasize the attack surface considerations. Microservices significantly increase the security complexity...",
      "created_at": "2025-10-30T13:23:37.000Z",
      "metadata": {
        "round": 1,
        "model": "claude-3-5-sonnet-20241022",
        "tokens": {
          "input": 480,
          "output": 385,
          "total": 865
        },
        "latency_ms": 1850
      }
    }
  ]
}

​
Message Structure

Messages are organized by debate rounds:
1

Round 1 - Initial Perspectives

Each agent provides their initial analysis of the question
2

Round 2 - Responses & Debate

Agents respond to each other’s points, building on or challenging ideas
3

Round 3 - Synthesis

Agents work toward consensus or clarify key disagreements
4

Additional Rounds

If configured, debates can have up to 5 rounds total

​
Analyzing Debates

​
Extract by Agent

const { messages } = await response.json();

// Group messages by agent
const byAgent = messages.reduce((acc, msg) => {
  if (!acc[msg.role]) acc[msg.role] = [];
  acc[msg.role].push(msg);
  return acc;
}, {});

console.log('CTO messages:', byAgent['CTO'].length);
console.log('Security Engineer messages:', byAgent['Security Engineer'].length);

​
Extract by Round

// Group messages by round
const byRound = messages.reduce((acc, msg) => {
  const round = msg.metadata?.round || 0;
  if (!acc[round]) acc[round] = [];
  acc[round].push(msg);
  return acc;
}, {});

console.log('Round 1 messages:', byRound[1].length);
console.log('Round 2 messages:', byRound[2].length);

​
Calculate Total Tokens

const totalTokens = messages.reduce((sum, msg) => {
  return sum + (msg.metadata?.tokens?.total || 0);
}, 0);

console.log(`Total tokens used: ${totalTokens}`);

​
Use Cases

Audit Trail

Keep complete records of AI reasoning for compliance

Fine-tuning

Use debate transcripts to train custom models

Quality Analysis

Analyze agent performance and consensus patterns

User Display

Show full conversation to end users in your app

​
Real-time Updates

To get messages as they’re created (instead of polling), use WebSocket: 
const ws = new WebSocket(task.websocket_url);

ws.on('message', (data) => {
  const event = JSON.parse(data);

  if (event.type === 'agent_message') {
    console.log(`${event.agent}: ${event.content}`);
  }

  if (event.type === 'round_started') {
    console.log(`Round ${event.round} started`);
  }

  if (event.type === 'task_completed') {
    console.log('Debate finished!');
    // Now fetch full messages for archival
  }
});
WebSockets give you real-time messages as the debate progresses. After completion, use this endpoint to get the complete history.

​
Message Ordering

Messages are returned in chronological order (oldest first):
  1. System messages (if any)
  2. Round 1 agent messages (in parallel, so timestamp determines order)
  3. Round 2 agent messages
  4. And so on…
The created_at timestamp reflects when each message was generated, not when it was sent to the client.

​
Error Responses

​
404
Not Found
Task ID doesn’t exist or is not an x402 task
​
500
Internal Server Error
Server error. Retry or contact support if persistent.

Create Task

Create a new debate task

Get Task

Check task status

List Rooms

Browse available rooms

​
Export Examples

​
Export to Markdown

function exportToMarkdown(messages) {
  let md = '# AI Council Debate\n\n';

  messages.forEach(msg => {
    md += `## ${msg.role}\n\n`;
    md += `${msg.content}\n\n`;
    md += `*Round ${msg.metadata.round} - ${msg.created_at}*\n\n`;
    md += '---\n\n';
  });

  return md;
}

​
Export to JSON

function exportToJSON(messages) {
  return JSON.stringify({
    debate_transcript: messages.map(msg => ({
      agent: msg.role,
      message: msg.content,
      round: msg.metadata.round,
      timestamp: msg.created_at,
      model: msg.metadata.model,
      tokens: msg.metadata.tokens
    }))
  }, null, 2);
}
Pro tip: Messages are immutable once created. Cache them locally to avoid repeated API calls!