Agent API

Prefix: /api/agent

Manages the background agent system — capability flags, task state, and health monitoring.

Flags

GET /api/agent/flags

Get current agent capability flags.

Response:

{
  "agent_enabled": false,
  "computer_use_enabled": false,
  "mcp_enabled": false,
  "browser_use_enabled": false
}

POST /api/agent/flags

Update agent flags. Changes are forwarded to the tool server.

Body:

{
  "lanlan_name": "character_name",
  "flags": {
    "agent_enabled": true,
    "mcp_enabled": true
  }
}

State & health

GET /api/agent/state

Get a snapshot of the agent's current state (running tasks, pending requests).

GET /api/agent/health

Agent health check endpoint.

Capability checks

GET /api/agent/computer_use/availability

Check if Computer Use is available (requires vision model configuration).

GET /api/agent/mcp/availability

Check if MCP (Model Context Protocol) is available.

GET /api/agent/user_plugin/availability

Check if user plugins are available.

GET /api/agent/browser_use/availability

Check if Browser Use is available.

Tasks

GET /api/agent/tasks

List all agent tasks (active and completed).

GET /api/agent/tasks/{task_id}

Get details for a specific task.

Commands

POST /api/agent/command

Send a control command to the agent.

Body:

{
  "lanlan_name": "character_name",
  "command": "pause",
  "task_id": "optional_task_id"
}

Available commands: pause, resume, cancel

Internal endpoints

POST /api/agent/internal/analyze_request

Internal endpoint for submitting analyze requests. Used by the main server's session manager.

POST /api/agent/admin/control

Admin control commands (e.g., kill process). Use with caution.