C3 AI Documentation Home

Web Search Tool

Use the web search tool to include real-time web data in Dynamic Agent responses.
You can configure it for specific instances of CanvasAgent_default or CanvasAgent_deep_research, connecting each to providers such as OpenAI Search API or Vertex AI Search API.

This document covers configuration for CanvasAgent_default and CanvasAgent_deep_research. All other agents are instances of Dynamic Agent Canvas or Dynamic Agent Deep Research, so you can follow the same steps for any agent instance.

Learn more about creating agents in Create Agents and Toolkits.

Prerequisites

Before you begin, ensure the following:

  • You can access Jupyter Notebook and the Application C3 AI Console.
  • LLM clients are pre-configured at the cluster level. Verify that web search-capable LLM clients are available in your environment.

Open Jupyter Notebook

Open Jupyter Notebook from the C3 Generative AI Application card:

  1. Navigate to your application in C3 AI Studio.
  2. Select Jupyter Notebook.
  3. Set the py-c3agents runtime as the notebook kernel.

Set up LLM credentials

LLM clients and authentication are pre-configured at the cluster level. Verify that your LLM clients are available by running the following command in the Application C3 AI Console:

Python
c3.GenaiCore.Llm.Completion.Client.listConfigKeys().collect()

For Vertex AI deployments, the default-completions-web client is pre-configured with Gemini 2.5 Flash for web search.

To configure additional LLM clients, see Set Up LLMs and Embedders Via UI or Set Up LLMs and Embedders Programmatically.

Get the list of agents

Get the config.name of the Dynamic Agent you want to configure the web search tool for. Run this command in your Python notebook.

Python
c3.Genai.Agent.Dynamic.Canvas.fetch()

List of agents

You can have multiple instances of CanvasAgent_default or CanvasAgent_deep_research. To configure the web search tool for a specific agent, identify the correct instance by checking the config.name value.

Web search tool for CanvasAgent_default

After you identify your agent, configure the web search tool for Dynamic Agent Canvas.

Get the current toolkit configuration

Python
toolkit = c3.Genai.Agent.Dynamic.Toolkit.forId('canvas_agent_default_toolkit')
print(f"Current toolkit: {toolkit.id}")
print(f"Current tools: {[tool.id for tool in toolkit.tools]}")
Python
# Get the pre-seeded web search tool
web_search_tool = c3.Genai.Agent.Dynamic.Tool.forId('openai_web_search')
print(f"Web search tool: {web_search_tool.id}")
print(f"Tool config params: {web_search_tool.toolConfigurationParams}")
Python
# Get current tools and add the web search tool
current_tools = toolkit.tools
updated_tools = current_tools.with_(web_search_tool)

# Update the toolkit with the new tool list
toolkit.withTools(updated_tools).merge()

print("OpenAI web search tool successfully added to toolkit")
Python
# Restart all engines to pick up the toolkit changes
c3.Genai.PyUtil.restartAllEngines()
print("Engines restarted successfully")

Web search tool for CanvasAgent_deep_research

Dynamic Agent Deep Research includes the web search tool out of the box and uses OpenAI Search API by default. You must set up OpenAI API credentials first. If you prefer Vertex AI, you can switch to it without needing credentials.

Dynamic Agent Deep Research works with OpenAI Search by default. After you set up the OpenAI API credentials in the first step, the agent can use the web search tool without additional configuration.

See also

Was this page helpful?