Trying the Web Search Agent: Setup and Execution with IO Intelligence

With io.net Intelligence, you can automate a wide range of tasks using pre-packaged AI Agents.

In this article, we focus on the Web Search Agent, and walk through everything from testing it in the UI to running it programmatically via the API.

What is the Web Search Agent?

As the name suggests, the Web Search Agent performs a basic search workflow:

• Searching for information on the web
• Summarizing the results and returning the output

On IO Intelligence, it can be used in the same way as other agents:

• Try it directly from the UI
• Integrate it into Python code
• Execute it via API using cURL

Opening the Web Search Agent from the Agents Page

[embed]io.net Deploy scalable clusters for AI development on a decentralized global GPU networkai.io.net

First, access IO Intelligence, then open the Agents tab from the navigation bar.

Steps:

• Navigate to the Agents tab
• Search for web search agent in the search bar
• Select the agent to open the details view

Testing in the Playground (No Installation Required)

On the agent details page, you will find a Try Agent button.

Clicking it opens the Playground, allowing you to test the agent before installing anything.

Procedure

• Click Try Agent
• Enter the topic you want to search for
• Run the agent by clicking Run Agent
• The output will appear in the Result tab

Since everything can be done through the UI, this is a convenient way to quickly validate the agent’s behavior.

Installing via pip

If you want to use the agent in a local environment, you can install it as a Python package:

pip install --upgrade iointel

This is the standard approach when embedding agents into your own applications.

Running the Web Search Agent from Python

After installation, the agent can be executed via the API.

Below is a minimal working example:

import os
import requests

api_key = os.environ["IO_API_KEY"]  # Replace with your actual io.net API Key
base_url = "https://api.intelligence.io.solutions/api/v1"

# Web Search Agent: A comprehensive web search agent that retrieves information from across the internet on any topic.
payload = {
    "objective": """Search the web for recent developments in quantum computing and summarize the top findings.""",
    "agent_names": ["web_search_agent"],
    "args": {
        "type": "custom",
        "name": "web-search",
        "objective": "Search and summarize web results",
        "instructions": "Focus on recent and credible sources"
    }
}

response = requests.post(
    f"{base_url}/workflows/run",
    headers={
        "Content-Type": "application/json",
        "Authorization": f"Bearer {api_key}"
    },
    json=payload
)

result = response.json()
print(result)

Key Parameters

objective

The main topic or question you want the agent to investigate.

instructions

Defines how the agent should search and summarize (e.g., prioritizing credible and recent sources).

Running the Agent Directly with cURL

You can also call the API directly without using Python:

curl -X POST \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $IO_API_KEY" \
  --data '{
    "objective": "Search the web for recent developments in quantum computing and summarize the top findings.",
    "agent_names": ["web_search_agent"],
    "args": {
        "type": "custom",
        "name": "web-search",
        "objective": "Search and summarize web results",
        "instructions": "Focus on recent and credible sources"
    }
}' \
  "https://api.intelligence.io.solutions/api/v1/workflows/run"

This is especially useful for quick testing or embedding into server-side workflows.

API Key Management (Inject API Key)

On the agent details page, IO Intelligence provides an Inject API Key button.

This allows you to:

• Generate a new API key
• Automatically insert it into the provided commands

This reduces setup friction during installation.

For production use, managing the key via environment variables is recommended:

export IO_API_KEY="xxxxx"

Secrets Management (When External Integrations Are Required)

Agents that integrate with third-party services such as GitHub or Linear require Secrets for authentication.

These can be configured in the Secret Management tab within the agent details view.

Information required for a Secret

• Secret Name
• Secret Value
• Associated Agent

Secrets can be added directly through the UI and will also appear in the API Keys and Secrets page.

Conclusion

In this article, we covered the basics of the Web Search Agent in io.net Intelligence:

• Trying it in the UI
• Installing via pip
• Running it through Python or cURL
• Managing API keys and Secrets

Because the Web Search Agent has a simple structure, it serves as a good starting point for understanding how agent execution works within IO Intelligence.