Chat API

AI conversation with streaming responses and agentic tool calling.

Endpoints

Method	Endpoint	Description
GET	`/api/agents/{id}/chat`	Get message history
POST	`/api/agents/{id}/chat`	Send message
DELETE	`/api/agents/{id}/chat`	Clear all history
GET	`/api/agents/{id}/chat/models`	List available models
GET	`/api/agents/{id}/chat/tools`	List available tools
GET	`/api/agents/{id}/chat/conversations`	List conversations
POST	`/api/agents/{id}/chat/conversations`	Create conversation
DELETE	`/api/agents/{id}/chat/conversations/{conversationId}`	Delete conversation
DELETE	`/api/agents/{id}/chat/messages/{messageId}`	Delete message
POST	`/api/agents/{id}/chat/messages/{messageId}/regenerate`	Regenerate response

Send Message

bash

POST /api/agents/{id}/chat
Content-Type: application/json

Request Body

json

{
  "content": "What's the weather like?",
  "conversationId": "conv-123",
  "model": "llama-4-scout",
  "enableTools": true,
  "autoRag": true,
  "stream": false
}

Request Parameters

Parameter	Type	Required	Description
`content`	string	Yes	The message content
`conversationId`	string	No	Continue existing conversation
`model`	string	No	Model to use (defaults to agent setting)
`enableTools`	boolean	No	Enable tool calling (default: true)
`autoRag`	boolean	No	Enable automatic RAG context (default: true)
`stream`	boolean	No	Enable streaming response (default: false)

Non-Streaming Response

json

{
  "userMessage": {
    "id": "msg_abc123",
    "role": "user",
    "content": "What's the weather like?",
    "conversationId": "conv-123",
    "createdAt": "2024-12-15T10:00:00Z"
  },
  "assistantMessage": {
    "id": "msg_def456",
    "role": "assistant",
    "content": "I don't have access to real-time weather data...",
    "conversationId": "conv-123",
    "toolCalls": null,
    "toolResults": null,
    "metadata": {
      "model": "llama-4-scout"
    },
    "createdAt": "2024-12-15T10:00:01Z"
  }
}

With Tool Calls

json

{
  "userMessage": { ... },
  "assistantMessage": {
    "id": "msg_def456",
    "role": "assistant",
    "content": "The weather in Tokyo is currently 22°C and sunny.",
    "toolCalls": [
      {
        "id": "tc_abc123",
        "name": "get_weather",
        "arguments": {"location": "Tokyo"}
      }
    ],
    "toolResults": [
      {
        "toolCallId": "tc_abc123",
        "name": "get_weather",
        "result": {"temperature": 22, "condition": "sunny"}
      }
    ]
  }
}

Streaming Response

For real-time streaming, set stream: true:

bash

POST /api/agents/{id}/chat
Content-Type: application/json

{
  "content": "Tell me a story",
  "stream": true
}

SSE Event Types

The streaming response uses Server-Sent Events with the following event types:

message_start

Sent when streaming begins.

event: message
data: {"type": "message_start", "userMessageId": "msg_abc123", "model": "llama-4-scout"}

thinking

Sent when the model is processing (for each agentic loop iteration).

event: message
data: {"type": "thinking", "iteration": 1}

tool_call

Sent when a tool is being called.

event: message
data: {"type": "tool_call", "id": "tc_abc123", "name": "search_knowledge", "arguments": {"query": "React tutorials"}}

tool_result

Sent when a tool execution completes.

event: message
data: {"type": "tool_result", "toolCallId": "tc_abc123", "name": "search_knowledge", "success": true, "result": [...], "resultPreview": "Found 5 results..."}

For errors:

event: message
data: {"type": "tool_result", "toolCallId": "tc_abc123", "name": "search_knowledge", "success": false, "error": "Search failed"}

content_delta

Sent for each chunk of the response text.

event: message
data: {"type": "content_delta", "delta": "Once upon a time", "index": 0}

message_complete

Sent when streaming finishes.

event: message
data: {"type": "message_complete", "assistantMessageId": "msg_def456", "toolCalls": [...], "toolResults": [...]}

error

Sent if an error occurs.

event: message
data: {"type": "error", "message": "Model request failed"}

Streaming Example

javascript

const response = await fetch('/api/agents/{id}/chat', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    content: 'Tell me a story',
    stream: true
  })
})

const reader = response.body.getReader()
const decoder = new TextDecoder()
let buffer = ''

while (true) {
  const { done, value } = await reader.read()
  if (done) break

  buffer += decoder.decode(value, { stream: true })
  const lines = buffer.split('\n')
  buffer = lines.pop() // Keep incomplete line in buffer

  for (const line of lines) {
    if (line.startsWith('data: ')) {
      const data = JSON.parse(line.slice(6))

      switch (data.type) {
        case 'message_start':
          console.log('Started:', data.model)
          break
        case 'thinking':
          console.log('Thinking... iteration', data.iteration)
          break
        case 'tool_call':
          console.log(`Calling: ${data.name}`)
          break
        case 'tool_result':
          if (data.success) {
            console.log(`Result: ${data.resultPreview}`)
          } else {
            console.log(`Error: ${data.error}`)
          }
          break
        case 'content_delta':
          process.stdout.write(data.delta)
          break
        case 'message_complete':
          console.log('\nDone!')
          break
        case 'error':
          console.error('Error:', data.message)
          break
      }
    }
  }
}

Aborting Streams

Use AbortController to cancel streaming:

javascript

const controller = new AbortController()

const response = await fetch('/api/agents/{id}/chat', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({ content: 'Tell me a story', stream: true }),
  signal: controller.signal
})

// Later, to abort:
controller.abort()

Get Messages

bash

GET /api/agents/{id}/chat?limit=50&conversationId=conv-123

Query Parameters

Parameter	Type	Description
`limit`	number	Max messages (default: 50)
`conversationId`	string	Filter by conversation
`before`	string	Messages before ID
`after`	string	Messages after ID

Response

json

{
  "messages": [
    {
      "id": "msg_abc123",
      "role": "user",
      "content": "Hello!",
      "conversationId": "conv-123",
      "createdAt": "2024-12-15T10:00:00Z"
    },
    {
      "id": "msg_def456",
      "role": "assistant",
      "content": "Hi there! How can I help you?",
      "conversationId": "conv-123",
      "toolCalls": null,
      "toolResults": null,
      "metadata": {
        "model": "llama-4-scout"
      },
      "createdAt": "2024-12-15T10:00:01Z"
    }
  ],
  "meta": {
    "total": 100,
    "hasMore": true
  }
}

Clear History

bash

DELETE /api/agents/{id}/chat

Query Parameters

Parameter	Type	Description
`conversationId`	string	Clear specific conversation only

Response

json

{
  "success": true,
  "deleted": 150
}

List Available Models

bash

GET /api/agents/{id}/chat/models

Response

json

{
  "models": [
    {
      "id": "llama-4-scout",
      "name": "Llama 4 Scout",
      "provider": "workers-ai",
      "contextWindow": 131072,
      "supportsTools": true,
      "supportsStreaming": true
    },
    {
      "id": "gpt-4o",
      "name": "GPT-4o",
      "provider": "openai",
      "contextWindow": 128000,
      "supportsTools": true,
      "supportsStreaming": true
    },
    {
      "id": "claude-3-5-sonnet",
      "name": "Claude 3.5 Sonnet",
      "provider": "anthropic",
      "contextWindow": 200000,
      "supportsTools": true,
      "supportsStreaming": true
    }
  ]
}

List Available Tools

bash

GET /api/agents/{id}/chat/tools

Response

json

{
  "tools": [
    {
      "name": "search_knowledge",
      "description": "Search the knowledge base for relevant information",
      "parameters": {
        "type": "object",
        "properties": {
          "query": {
            "type": "string",
            "description": "Search query"
          }
        },
        "required": ["query"]
      }
    },
    {
      "name": "browse_url",
      "description": "Fetch and read content from a URL",
      "parameters": {
        "type": "object",
        "properties": {
          "url": {
            "type": "string",
            "description": "URL to browse"
          }
        },
        "required": ["url"]
      }
    },
    {
      "name": "execute_sql",
      "description": "Execute a SQL query against the database",
      "parameters": {
        "type": "object",
        "properties": {
          "query": {
            "type": "string",
            "description": "SQL query to execute"
          }
        },
        "required": ["query"]
      }
    }
  ]
}

Conversations API

List Conversations

bash

GET /api/agents/{id}/chat/conversations

Response

json

{
  "conversations": [
    {
      "id": "conv-123",
      "title": "Product Questions",
      "messageCount": 15,
      "lastMessageAt": "2024-12-15T10:30:00Z",
      "createdAt": "2024-12-15T10:00:00Z"
    },
    {
      "id": "conv-456",
      "title": "Technical Support",
      "messageCount": 8,
      "lastMessageAt": "2024-12-14T15:20:00Z",
      "createdAt": "2024-12-14T15:00:00Z"
    }
  ]
}

Create Conversation

bash

POST /api/agents/{id}/chat/conversations
Content-Type: application/json

{
  "title": "New Chat"
}

Response

json

{
  "id": "conv-789",
  "title": "New Chat",
  "messageCount": 0,
  "createdAt": "2024-12-15T11:00:00Z"
}

Delete Conversation

bash

DELETE /api/agents/{id}/chat/conversations/{conversationId}

Response

json

{
  "success": true,
  "deleted": 1
}

Message Management

Delete Message

bash

DELETE /api/agents/{id}/chat/messages/{messageId}

Response

json

{
  "success": true
}

Regenerate Response

Regenerate the assistant's response for a message.

bash

POST /api/agents/{id}/chat/messages/{messageId}/regenerate
Content-Type: application/json

{
  "model": "gpt-4o",
  "enableTools": true,
  "stream": true
}

The response format matches the standard message response (streaming or non-streaming based on the stream parameter).

Message Format

ChatMessage

typescript

interface ChatMessage {
  id: string
  role: "user" | "assistant" | "system" | "tool"
  content: string
  conversationId?: string
  toolCalls?: ToolCall[]
  toolResults?: ToolResult[]
  metadata?: {
    model?: string
    toolCalls?: ToolCall[]
    toolResults?: ToolResult[]
  }
  createdAt: string
}

ToolCall

typescript

interface ToolCall {
  id: string          // 9-character alphanumeric ID
  name: string        // Tool name
  arguments: Record<string, unknown>  // Tool parameters
}

ToolResult

typescript

interface ToolResult {
  toolCallId: string  // References ToolCall.id
  name: string        // Tool name
  result: unknown     // Tool output
  error?: string      // Error message if failed
}

Errors

Code	Description
400	Invalid message or parameters
404	Conversation or message not found
409	Conversation locked (concurrent request)
429	Rate limited
500	AI service error

Error Response

json

{
  "error": "Conversation is currently being processed",
  "code": "CONVERSATION_LOCKED"
}

Context Window Management

The API automatically manages context:

Token estimation: ~4 characters = 1 token
Reserved tokens: ~20K for system prompt, tools, RAG, and response
Automatic trimming: Older messages removed when approaching limits
System prompt: Always included in context

Agentic Loop

When tools are enabled, the API uses an agentic loop:

Maximum 8 tool call iterations per message
Loop detection prevents repeated calls with same arguments
Tool results are added to conversation for context
Large results (like screenshots) are truncated

Examples

Simple Chat

bash

curl -X POST https://your-domain.com/api/agents/{id}/chat \
  -H "Content-Type: application/json" \
  -d '{"content": "Hello!"}'

Chat with Model Selection

bash

curl -X POST https://your-domain.com/api/agents/{id}/chat \
  -H "Content-Type: application/json" \
  -d '{
    "content": "Explain quantum computing",
    "model": "gpt-4o"
  }'

Streaming with Tools

bash

curl -X POST https://your-domain.com/api/agents/{id}/chat \
  -H "Content-Type: application/json" \
  -d '{
    "content": "Search for recent news about AI",
    "enableTools": true,
    "stream": true
  }'

Continue Conversation

bash

curl -X POST https://your-domain.com/api/agents/{id}/chat \
  -H "Content-Type: application/json" \
  -d '{
    "content": "Tell me more about that",
    "conversationId": "conv-123"
  }'

Chat API ​

Endpoints ​

Send Message ​

Request Body ​

Request Parameters ​

Non-Streaming Response ​

With Tool Calls ​

Streaming Response ​

SSE Event Types ​

message_start ​

thinking ​

tool_call ​

tool_result ​

content_delta ​

message_complete ​

error ​

Streaming Example ​

Aborting Streams ​

Get Messages ​

Query Parameters ​

Response ​

Clear History ​

Query Parameters ​

Response ​

List Available Models ​

Response ​

List Available Tools ​

Response ​

Conversations API ​

List Conversations ​

Response ​

Create Conversation ​

Response ​

Delete Conversation ​

Response ​

Message Management ​

Delete Message ​

Response ​

Regenerate Response ​

Message Format ​

ChatMessage ​

ToolCall ​

ToolResult ​

Errors ​

Error Response ​

Context Window Management ​

Agentic Loop ​

Examples ​

Simple Chat ​

Chat with Model Selection ​

Streaming with Tools ​

Continue Conversation ​

Chat API

Endpoints

Send Message

Request Body

Request Parameters

Non-Streaming Response

With Tool Calls

Streaming Response

SSE Event Types

message_start

thinking

tool_call

tool_result

content_delta

message_complete

error

Streaming Example

Aborting Streams

Get Messages

Query Parameters

Response

Clear History

Query Parameters

Response

List Available Models

Response

List Available Tools

Response

Conversations API

List Conversations

Response

Create Conversation

Response

Delete Conversation

Response

Message Management

Delete Message

Response

Regenerate Response

Message Format

ChatMessage

ToolCall

ToolResult

Errors

Error Response

Context Window Management

Agentic Loop

Examples

Simple Chat

Chat with Model Selection

Streaming with Tools

Continue Conversation