Skip to main content

​
Sub-Agents System

Complexity in AI tasks often requires specialization. The Sub-Agents System allows you to build a hierarchical architecture where a “Parent” agent can intelligently delegate tasks to specialized “Child” agents.

​
Why Use Sub-Agents?

  • Modular Design: Build and test small, expert agents individually before integrating them.
  • Improved Accuracy: Specialized agents (e.g., a “Math Expert” or “Code Auditor”) are less likely to hallucinate in their specific domain.
  • Tool Management: Avoid cluttering a single agent’s prompt with hundreds of tools. Instead, give each sub-agent only the tools it needs.

​
Configuration

To use sub-agents, create them individually and then pass them as a list to the parent agent. 
from omnicoreagent import OmniCoreAgent

# 1. Create specialized child agents
research_agent = OmniCoreAgent(
    name="researcher",
    system_instruction="You specialize in web research and summarization.",
    ...
)

code_agent = OmniCoreAgent(
    name="coder",
    system_instruction="You are an expert Python developer.",
    ...
)

# 2. Integrate them into the parent agent
parent_agent = OmniCoreAgent(
    name="manager",
    system_instruction="You manage a team of specialists to complete complex requests.",
    sub_agents=[research_agent, code_agent],
    ...
)

​
How Delegation Works

When a parent agent has sub-agents, it is automatically given tools to communicate with them:
ToolAction
call_sub_agentSend a specific task to a child agent and wait for the result.
The parent agent decides when a task is too complex or specific and which sub-agent is best suited to handle it. The conversation history of the sub-agent is managed independently but can be accessed by the parent.

​
Example Workflow

  1. User Query: “Research the latest AI news and write a Python script for a news aggregator.”
  2. Manager (Parent): Recognizes this requires research and coding.
  3. Manager: Calls research_agent with the task “Get latest AI news summaries”.
  4. Researcher (Sub): Performs web search, returns summaries.
  5. Manager: Receives summaries, then calls code_agent with “Write a Python aggregator using this data: [summaries]”.
  6. Coder (Sub): Writes the code, returns it.
  7. Manager: Synthesizes the final response to the user.

​
Best Practices

  • Explicit Instructions: In the parent agent’s system instruction, briefly describe the expertise of each sub-agent so it knows who to turn to.
  • Limit Depth: While hierarchical agents are powerful, avoid nesting agents too deep (e.g., Parent -> Child -> Grandchild) to prevent excessive latency and token usage.
  • Shared Session IDs: Use the same session_id for both parent and child calls if you want them to share context indirectly through the memory store.