Build Your First Agent¶
This guide walks you through creating an agent with a tool and full tracing in under 5 minutes.
Prerequisites¶
- Python 3.10+
- An OpenAI API key (set as
OPENAI_API_KEYenvironment variable)
Step 1: Install¶
Step 2: Create an Agent with a Tool¶
from fastaiagent import Agent, LLMClient
from fastaiagent.tool import FunctionTool
# Define a tool
weather_tool = FunctionTool(
name="get-weather",
description="Get current weather for a city",
fn=lambda city: {"city": city, "temp": "72F", "condition": "sunny"}
)
# Create an agent
agent = Agent(
name="assistant",
system_prompt="You are a helpful assistant. Use tools when needed.",
llm=LLMClient(provider="openai", model="gpt-4o-mini"),
tools=[weather_tool]
)
# Run it — every run is automatically traced
result = agent.run("What's the weather in San Francisco?")
print(result.output)
print(result.trace_id) # e.g. "b6acf1ef2c2779bbc2fcf80802ae0534"
Step 3: View and Replay the Trace¶
from fastaiagent.trace import Replay
# Load the trace using the trace_id from the result
replay = Replay.load(result.trace_id)
print(replay.summary())
# Trace: b6acf1ef2c2779bbc2fcf80802ae0534
# Steps: 3 | Duration: 1.2s | Tokens: 245
# Step 1: LLM call (choose tool) - 400ms
# Step 2: Tool call (get-weather) - 5ms
# Step 3: LLM call (final response) - 750ms
Step 4: Browse Past Traces¶
from fastaiagent.trace import TraceStore
store = TraceStore()
for t in store.list_traces(last_hours=24):
print(f"{t.trace_id[:12]} {t.name} {t.status}")
Or via CLI:
That's it. You've built an agent with a tool and full tracing.
Next Steps¶
- Add guardrails to validate inputs and outputs
- Build a chain workflow with loops and checkpointing
- Debug with Agent Replay using fork-and-rerun
- Connect to the platform for visual editing
- Evaluate your agent with scorers and datasets