Add Web Search to CrewAI Agents in 5 Minutes

The Problem

CrewAI agents are powerful for multi-step reasoning, but they're limited to their training data. A research agent that can't search the web is like a librarian locked out of the library.

Setup

bash
pip install crewai httpx

Get a free API key from Searlo Dashboard (includes 3,000 credits).

Create the Search Tool

python
import httpx
from crewai.tools import BaseTool
from pydantic import Field

class WebSearchTool(BaseTool):
    name: str = "Web Search"
    description: str = "Search Google for real-time information. Use for current events, facts, or research."
    api_key: str = Field(default="your_searlo_api_key")

    def _run(self, query: str) -> str:
        response = httpx.get(
            "https://api.searlo.tech/api/v1/search/web",
            params={"q": query, "limit": 5, "format": "toon"},
            headers={"X-API-Key": self.api_key},
            timeout=10.0,
        )
        data = response.json()
        results = data.get("organic", [])
        if not results:
            return "No results found."
        return "\n\n".join(
            f"**{r['title']}**\n{r['snippet']}\nSource: {r['link']}"
            for r in results
        )

Use It in a Crew

python
from crewai import Agent, Task, Crew

search = WebSearchTool(api_key="your_searlo_api_key")

researcher = Agent(
    role="Senior Research Analyst",
    goal="Find accurate, current information on the given topic",
    backstory="Expert researcher who always verifies facts with multiple sources.",
    tools=[search],
    verbose=True,
)

writer = Agent(
    role="Content Writer",
    goal="Write clear, well-sourced articles based on research",
    backstory="Experienced writer who transforms research into readable content.",
    verbose=True,
)

research_task = Task(
    description="Research the current state of {topic}. Find at least 3 recent sources.",
    expected_output="A research brief with key findings and source URLs.",
    agent=researcher,
)

write_task = Task(
    description="Write a 500-word article based on the research brief.",
    expected_output="A well-structured article with source citations.",
    agent=writer,
)

crew = Crew(agents=[researcher, writer], tasks=[research_task, write_task])
result = crew.kickoff(inputs={"topic": "AI regulation in the EU"})
print(result)

What Happens

1. The researcher agent searches Google via Searlo for current information

2. It can make multiple search calls to verify facts across sources

3. The writer agent receives the research and creates the article

4. Total cost: ~5-10 searches = less than $0.01

Why Searlo for CrewAI

• MCP server: the same key works as a search tool in Claude, Cursor and any other MCP client

• TOON format: 63% fewer tokens means cheaper LLM calls in the pipeline

• No subscription: One-time credits work well for batch agent runs

• Structured JSON: Clean data that agents can parse reliably

More Resources

• Full integrations guide with LangChain, LlamaIndex, and n8n

• Search API for AI Agents — architecture patterns and use cases

• Pricing — credit packs start at $3.99