Skip to main content
POST
/
api
/
v1
/
agent
/
chat
Chat
curl --request POST \
  --url https://api.agent-mind.com/api/v1/agent/chat \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "message": "<string>",
  "conversation_id": "<string>",
  "user_id": "<string>",
  "session_id": "<string>",
  "context": {},
  "stream": true
}
'

Documentation Index

Fetch the complete documentation index at: https://docs.agent-mind.com/llms.txt

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

​
Overview

Chat with your configured LeedAB agent programmatically. Your agent has access to all your memories, custom tools, and web search capabilities. Perfect for building chatbots, customer support systems, or any application that needs intelligent AI responses. Key Features:
  • Access to all your memories (with user_id segmentation for multi-tenant apps)
  • All your custom tools available
  • Web search and webpage reading built-in
  • Streaming (SSE) or synchronous responses
  • Persistent conversation history

​
Request Body

​
message
string
required
The message to send to the agent.
​
conversation_id
string
Continue an existing conversation. If not provided, a new conversation is created.
​
user_id
string
Important for multi-tenant apps: Segment memories by end-user. When provided, the agent’s remember, recall, and forget tools only access memories tagged with this user_id. This lets you serve multiple end-users while keeping their data separate.
​
session_id
string
Optional session identifier for grouping related conversations. Useful for filtering conversations later.
​
context
object
Additional context to include with the message. Key-value pairs that will be prepended to the message.
​
stream
boolean
default:"true"
Whether to stream the response (Server-Sent Events) or return the complete response. Default is streaming.

​
Response

​
Non-Streaming Response (stream: false)

Returns the complete agent response as JSON. 
{
  "response": "Based on your previous meetings, here's a summary...",
  "conversation_id": "550e8400-e29b-41d4-a716-446655440000",
  "message_id": "660e8400-e29b-41d4-a716-446655440001",
  "user_id": "customer_123",
  "credits_remaining": 94,
  "credits_used": 1
}

​
Streaming Response (stream: true)

Returns Server-Sent Events (SSE) with real-time agent output. 
data: {"type": "thinking", "content": "", "metadata": {"conversation_id": "...", "message_id": "..."}}

data: {"type": "thinking_step", "content": "Let me search my memories...", "metadata": {"step_number": 1}}

data: {"type": "tool_start", "content": "", "metadata": {"tool_name": "recall", "arguments": {"query": "meetings"}}}

data: {"type": "tool_end", "content": "Found 3 relevant memories...", "metadata": {"step_number": 2}}

data: {"type": "complete", "content": "Based on your previous meetings, here's a summary..."}

data: [DONE]

​
Examples

​
Basic Chat

import requests

url = "https://api.leedab.com/api/v1/agent/chat"
headers = {
    "Authorization": "Bearer am_live_your_api_key",
    "Content-Type": "application/json"
}
data = {
    "message": "What do you know about my project?",
    "stream": False
}

response = requests.post(url, json=data, headers=headers)
result = response.json()
print(result["response"])

​
Multi-Tenant App (User Segmentation)

Use user_id to give each of your end-users their own memory space: 
# Each customer gets their own segmented memories
data = {
    "message": "Remember that I prefer dark mode",
    "user_id": "customer_123",  # This customer's memories are isolated
    "stream": False
}
response = requests.post(url, json=data, headers=headers)

# Later, when this customer returns
data = {
    "message": "What are my preferences?",
    "user_id": "customer_123",  # Only sees their own memories
    "stream": False
}
response = requests.post(url, json=data, headers=headers)

​
Streaming Response

import requests

url = "https://api.leedab.com/api/v1/agent/chat"
headers = {
    "Authorization": "Bearer am_live_your_api_key",
    "Content-Type": "application/json"
}
data = {
    "message": "Summarize my recent meetings",
    "stream": True  # Default
}

# Stream the response
response = requests.post(url, json=data, headers=headers, stream=True)
for line in response.iter_lines():
    if line:
        line = line.decode('utf-8')
        if line.startswith('data: '):
            event_data = line[6:]  # Remove 'data: ' prefix
            if event_data == '[DONE]':
                break
            print(event_data)

​
Continue Conversation

# First message creates a conversation
data = {"message": "Hi, I'm working on a Python project", "stream": False}
response = requests.post(url, json=data, headers=headers)
conversation_id = response.json()["conversation_id"]

# Continue the same conversation
data = {
    "message": "What libraries should I use for web scraping?",
    "conversation_id": conversation_id,
    "stream": False
}
response = requests.post(url, json=data, headers=headers)

​
With External Context

data = {
    "message": "Help me debug this issue",
    "context": {
        "current_page": "/checkout",
        "cart_total": "$149.99",
        "error_code": "PAYMENT_FAILED"
    },
    "user_id": "customer_456",
    "stream": False
}
response = requests.post(url, json=data, headers=headers)

​
Streaming Event Types

When stream: true, you’ll receive these event types:
TypeDescription
thinkingInitial event with conversation/message IDs
thinking_stepAgent’s reasoning process
planningAgent’s plan for the task
code_actionPython code being executed
tool_startTool invocation starting
tool_endTool result
completeFinal agent response
errorError message if something went wrong

​
Error Responses

​
402 Payment Required

{
  "error": "insufficient_credits",
  "message": "You've run out of credits. Purchase more at https://app.leedab.com/billing",
  "balance": 0,
  "total_purchased": 100
}

​
404 Not Found

{
  "error": "conversation_not_found",
  "message": "Conversation not found"
}

​
500 Internal Server Error

{
  "error": "agent_error",
  "message": "Error details here"
}