Skip to main content

​
Overview

The agent_configs table stores configuration data for individual agents within workflows. Each agent element in a workflow can have its own configuration settings stored in this table.

​
Table Schema

​
id
string
required
Unique identifier for the configuration record (auto-generated UUID)
​
user_id
string
required
Foreign key reference to the authenticated user
​
element_id
string
required
Identifier of the workflow element this configuration belongs to
​
agent_type
string
required
Type of agent (e.g., “processor”, “transformer”, “analyzer”)
​
config
object
required
Agent-specific configuration data stored as JSON. The structure varies by agent type.
​
created_at
string
ISO 8601 timestamp of when the configuration was created
​
updated_at
string
ISO 8601 timestamp of the last configuration update

​
TypeScript Interface

export interface AgentConfig {
  id: string;
  user_id: string;
  element_id: string;
  agent_type: string;
  config: Record<string, any>;
  created_at?: string;
  updated_at?: string;
}

// Agent output structure for configuration
export interface AgentOutputStructure {
  type: string;
  fields: AgentOutputField[];
  description?: string;
}

export interface AgentOutputField {
  name: string;
  type: 'string' | 'number' | 'boolean' | 'object' | 'array';
  description?: string;
  required?: boolean;
}

​
CRUD Operations

​
Save Configuration

Saves or updates an agent’s configuration. Uses upsert to either create a new configuration or update an existing one based on the user_id and element_id combination. Endpoint: manage-agent-configs Request: 
{
  "action": "save-config",
  "elementId": "element-123",
  "agentType": "data-processor",
  "config": {
    "apiEndpoint": "https://api.example.com",
    "timeout": 30000,
    "retries": 3,
    "settings": {
      "enableLogging": true,
      "logLevel": "info"
    }
  }
}
Response: 
{
  "success": true
}
Error Response: 
{
  "error": "Element ID and agent type are required"
}

​
Get Configuration

Retrieves the saved configuration for a specific agent element. Endpoint: manage-agent-configs Request: 
{
  "action": "get-config",
  "elementId": "element-123",
  "agentType": "data-processor"
}
Response: 
{
  "config": {
    "apiEndpoint": "https://api.example.com",
    "timeout": 30000,
    "retries": 3,
    "settings": {
      "enableLogging": true,
      "logLevel": "info"
    }
  }
}
If no configuration exists, returns an empty config: 
{
  "config": {}
}

​
Relationships

  • user_id → References the authenticated user’s ID
  • element_id → References a workflow element ID from the user_workflows.data.elements array
  • Configuration is linked to specific workflow elements, allowing each agent instance to have unique settings

​
Constraints

  • Unique constraint: The combination of user_id and element_id must be unique (enforced by upsert with onConflict: 'user_id,element_id')
  • The element_id should correspond to an element in the user’s workflow
  • The config field must contain valid JSON
  • Both element_id and agent_type are required for all operations

​
Usage Notes

  • The config object structure is flexible and varies by agent type
  • Configurations are persisted separately from workflow data for better modularity
  • When an agent element is duplicated in a workflow, it should reference a different element_id to maintain separate configurations
  • The upsert operation ensures that saving a configuration multiple times updates the existing record rather than creating duplicates

Build docs developers (and LLMs) love

Get started for freeTalk to us