API Reference

POST /api/chat

Chat with the agent. Supports streaming (SSE) and non-streaming responses.

Request Body

{
  "messages": [{ "role": "user", "content": "Search for our deployment docs" }],
  "userId": "user-123",
  "chatId": "chat-456",
  "model": "gpt-4o",
  "temperature": 0.7,
  "stream": false,
  "enableMemory": true,
  "tools": ["search", "file_access"]
}

Field	Type	Required	Default	Description
`messages`	`Array`	Yes	—	Chat messages
`userId`	`string`	Yes	—	User identifier
`chatId`	`string`	No	auto-generated	Conversation identifier
`model`	`string`	No	`gpt-4o`	LLM model
`stream`	`boolean`	No	`false`	Enable SSE streaming
`enableMemory`	`boolean`	No	`true`	Load/store conversation memory
`tools`	`string[]`	No	—	Restrict available tools

Response (Non-Streaming)

{
  "message": {
    "role": "assistant",
    "content": "I found the deployment documentation...",
    "timestamp": "2026-02-21T12:00:00Z"
  },
  "chatId": "chat-456",
  "toolsUsed": ["search"],
  "memoryUpdated": true
}

Streaming Events

Event	Data	Description
`token`	`string`	Response token chunk
`tool_start`	`{tool, parameters}`	Tool execution started
`tool_result`	`{tool, success, data, error}`	Tool execution result
`done`	`{chatId, toolsUsed, memoryUpdated}`	Stream complete
`error`	`{error}`	Error occurred

GET /api/chat/:chatId

Retrieve full conversation history.

Query Parameters

Parameter	Required	Description
`userId`	Yes	User identifier

Response

{
  "chatId": "chat-456",
  "userId": "user-123",
  "messages": [...],
  "lastActivity": "2026-02-21T12:00:00Z",
  "messageCount": 10
}

Memory Endpoints

GET /api/memory/:userId

List memories. Query params: type (default: conversation), chatId, limit (default: 50).

POST /api/memory

Store a memory.

{
  "userId": "user-123",
  "type": "long_term",
  "content": "User prefers Python code examples",
  "importance": 8,
  "tags": ["preferences", "coding"]
}

POST /api/memory/:userId/search

Search memories by text query.

{ "query": "coding preferences", "type": "long_term", "limit": 10 }

GET /api/memory/:userId/summary

Generate a text summary of memories. Query params: type, chatId.

DELETE /api/memory/:userId/conversation/:chatId

Delete all memories for a conversation.

DELETE /api/memory/item/:memoryId

Delete a single memory by ID.

Task Endpoints (v1 mode only)

GET /api/tasks

List tasks. Query params: userId (required), status, limit.

POST /api/tasks

Create a task.

{
  "title": "Analyze competitor landscape",
  "description": "Research and compare top 5 competitors",
  "userId": "user-123",
  "chatId": "chat-456",
  "steps": [
    { "title": "Research", "description": "Find competitor info", "dependencies": [] },
    { "title": "Compare", "description": "Create comparison matrix", "dependencies": ["Research"] },
    { "title": "Report", "description": "Write final report", "dependencies": ["Compare"] }
  ]
}

GET /api/tasks/:taskId

Full task status with agents, steps, and results.

POST /api/tasks/:taskId/execute

Start execution (returns 202 Accepted). Poll GET /api/tasks/:taskId for status.

DELETE /api/tasks/:taskId

Cancel a pending or running task.

Tool Endpoints

GET /api/tools

List all tools with their schemas and enabled status.

GET /api/tools/:toolName

Get details for a specific tool.

POST /api/tools/execute

Execute a tool directly.

{
  "toolName": "search",
  "parameters": { "query": "deployment guide" },
  "userId": "user-123"
}

PATCH /api/tools/:toolName

Enable or disable a tool.

{ "enabled": false }

Configuration Overview