api_database-api/.trae/documents/Implement Dynamic Agent API with Video and Knowledge Base Support.md

Add Dynamic Agent API (Type=1) with Database Integration

I will implement the Dynamic Agent feature by integrating it with the existing agent_cards system. Dynamic Agents will be distinguished by type=1 and will store their extended data (videos, knowledge base) in a new linked table.

1. Database Schema Extension

I will create a new table dynamic_agent_details to store the specialized data for dynamic agents.

• Table Name: dynamic_agent_details
• Columns:
  • id: SERIAL PRIMARY KEY
  • card_id: INTEGER (Foreign Key linking to agent_cards.card_id)
  • videos: JSONB (Stores list of {url, emotion}, min 1, max 7)
  • kb_id: VARCHAR (Knowledge Base ID)
  • kb_config: JSONB (Additional Knowledge Base fields)
  • created_at: TIMESTAMP

2. Code Implementation (main.py)

Pydantic Models

I will define:

• VideoItem: {url: str, emotion: str}
• DynamicAgentCreate: Inherits/Includes fields from AgentCard (name, avatar, etc.) PLUS:
  • videos: List[VideoItem] (validated length 1-7)
  • kb_id: str
  • kb_config: dict

API Endpoints

1. Create Dynamic Agent

   • POST /dynamic_agent/
   • Logic:
     1. Insert the base agent info into agent_cards with type=1 and RETURNING card_id.
     2. Insert the extended info (videos, kb details) into dynamic_agent_details using the returned card_id.
     3. Return the new card_id.

2. Get Dynamic Agent Details

   • GET /dynamic_agent/{card_id}
   • Logic:
     1. Fetch base info from agent_cards.
     2. Fetch extended info from dynamic_agent_details.
     3. Return combined result.

Helper Functions

• init_dynamic_agent_db(): Checks/Creates the dynamic_agent_details table on startup.

3. Verification

• Create a dynamic agent via the new API.
• Verify it appears in the agent_cards table (as type 1).
• Verify extended data is in dynamic_agent_details.