agenticx-agent-builder

AgenticX Agent Builder

Guide for creating production-grade agents in AgenticX.

Core Concepts

An Agent in AgenticX consists of:

• Identity: id, name, role, goal
• LLM Provider: the language model backend
• Tools: functions the agent can invoke
• Executor: the runtime that orchestrates agent reasoning

Creating an Agent

Minimal Agent

from agenticx import Agent, Task, AgentExecutor
from agenticx.llms import OpenAIProvider

agent = Agent(
    id="assistant",
    name="Assistant",
    role="General Purpose Assistant",
    goal="Help users with tasks",
    organization_id="default"
)

Agent with Rich Configuration

agent = Agent(
    id="research-analyst",
    name="Research Analyst",
    role="Senior Research Analyst",
    goal="Produce thorough, well-cited research reports",
    backstory="10 years experience in data-driven research",
    organization_id="research-team",
    verbose=True
)

CLI Creation

agx agent create research-analyst --role "Senior Research Analyst"
agx agent list

LLM Providers

AgenticX supports multiple LLM backends through a unified interface:

from agenticx.llms import OpenAIProvider, LiteLLMProvider

# OpenAI
llm = OpenAIProvider(model="gpt-4")

# Any model via LiteLLM (Codex, Gemini, local models, etc.)
llm = LiteLLMProvider(model="anthropic/Codex-sonnet-4-20250514")
llm = LiteLLMProvider(model="ollama/llama3")

Adding Tools

Function Decorator Tools

from agenticx.tools import tool

@tool
def search_web(query: str) -> str:
    """Search the web for information."""
    return f"Results for: {query}"

@tool
def calculate(expression: str) -> float:
    """Evaluate a math expression safely."""
    return eval(expression)  # use ast.literal_eval in production

Attaching Tools to Execution

executor = AgentExecutor(
    agent=agent,
    llm=llm,
    tools=[search_web, calculate]
)
result = executor.run(task)

Task Definition