smolagents

Hugging Face developed smolagents as a minimalist, open-source framework for building AI agents with minimal code. The library prioritizes simplicity and efficiency, enabling developers to deploy LLM-powered agents quickly.

This guide covers capturing telemetry from smolagents applications using InteractiveAI.

Prerequisites

• InteractiveAI account with API credentials

• Hugging Face token

---

Installation

pip install interactiveai 'smolagents[telemetry]' opentelemetry-sdk opentelemetry-exporter-otlp openinference-instrumentation-smolagents

---

Configuration

Set your API credentials as environment variables:

import os

# InteractiveAI credentials
# Obtain keys from Settings > API Keys in the dashboard
os.environ["INTERACTIVEAI_PUBLIC_KEY"] = "pk-ia-..."
os.environ["INTERACTIVEAI_SECRET_KEY"] = "sk-ia-..."

# Hugging Face token
os.environ["HF_TOKEN"] = "hf_..."

Initialize the client and confirm connectivity:

from interactiveai import Interactive

client = Interactive(
    public_key=os.environ["INTERACTIVEAI_PUBLIC_KEY"],
    secret_key=os.environ["INTERACTIVEAI_SECRET_KEY"],
    host=os.environ.get("INTERACTIVEAI_HOST", "https://app.interactiveai.com")
)

if client.auth_check():
    print("Connection established")
else:
    print("Authentication failed - verify credentials")

---

Enabling Trace Capture

Initialize the SmolagentsInstrumentor before your application code:

from openinference.instrumentation.smolagents import SmolagentsInstrumentor

SmolagentsInstrumentor().instrument()

---

Running a smolagent

Here's an example with a manager agent orchestrating a search agent:

import os
from smolagents import (
    CodeAgent,
    ToolCallingAgent,
    DuckDuckGoSearchTool,
    VisitWebpageTool,
    OpenAIModel,
)

model = OpenAIModel(model_id="gpt-4o-mini")

search_agent = ToolCallingAgent(
    tools=[DuckDuckGoSearchTool(), VisitWebpageTool()],
    model=model,
    name="search_agent",
    description="This agent performs web searches and retrieves webpage content.",
)

manager_agent = CodeAgent(
    tools=[],
    model=model,
    managed_agents=[search_agent],
)

manager_agent.run(
    "Research the current adoption rates of renewable energy in Europe and summarize the key trends."
)

---

Enriching Traces with Context

Combine the SmolagentsInstrumentor with the InteractiveAI SDK to attach identifiers and metadata:

---

Trace Visibility

The InteractiveAI dashboard displays:

• Manager and managed agent interactions

• Tool invocations with search queries and results

• LLM calls with prompts and completions

• Multi-step reasoning chains

• Token consumption and latency metrics