Using Tools

What are Tools?

Tools give your agents the ability to take real actions - like calculating numbers, reading files, scraping websites, or calling APIs. LangCrew’s tool system automatically finds and provides access to tools from multiple sources.

Quick example:

from langcrew.tools import ToolRegistry

# Get a tool and use it
calculator = ToolRegistry.get_tool("calculator")
result = calculator.run("2 + 2")  # "4"

Getting Tools: Three Ways

1. Install Tool Packages (Recommended)

Best for: Most users - gets you hundreds of ready-to-use tools

pip install crewai_tools langchain_community

Popular tools you’ll get:

from langcrew.tools import ToolRegistry

# CrewAI tools (from crewai_tools package)
file_read = ToolRegistry.get_tool("crewai:read_a_files_content")
scraper = ToolRegistry.get_tool("crewai:read_website_content")
search = ToolRegistry.get_tool("crewai:search_the_internet_with_serper")
# ... many more CrewAI tools available

# LangChain Community tools (from langchain_community package)
# Note: These require additional dependencies
wikipedia = ToolRegistry.get_tool("langchain:wikipedia")
calculator = ToolRegistry.get_tool("langchain:calculator")
# ... many more LangChain tools available

Important: Tool names use provider prefixes and snake_case format:

• crewai: prefix for CrewAI tools (from crewai_tools) - use snake_case format
• langchain: prefix for LangChain community tools (from langchain_community.tools) - use snake_case format
• langcrew: prefix for LangCrew-specific tools (from langcrew_tools)
• No prefix for custom tools in your project

Note: Always use snake_case format (lowercase with underscores) for tool names, even if the original tool uses descriptive names with spaces.

To see all available tools:

python -c "from langcrew.tools import ToolRegistry; print(ToolRegistry.list_tools())"

Note: External tools are loaded on-demand (lazy loading) - they only load when first requested to avoid import errors from missing dependencies.

2. Create Custom Tools

Best for: Business-specific logic or unique workflows

Create a tools/ directory in your project:

tools/my_tools.py

from langchain_core.tools import BaseTool

class SalesReportTool(BaseTool):
    name = "sales_report"
    description = "Generate sales reports"

    def _run(self, period: str = "monthly") -> str:
        # Your business logic here
        return f"Sales for {period}: $50,000"

Use it immediately:

report_tool = ToolRegistry.get_tool("sales_report")
result = report_tool.run("quarterly")

3. Manual Registration

Best for: Advanced users who need to override existing tools

from langcrew.tools import ToolRegistry

# Register your own version
ToolRegistry.register("calculator", MyEnhancedCalculator())

# Now your version is used
calc = ToolRegistry.get_tool("calculator")

Using Tools with Agents

Here’s a complete example showing tools in action:

from langcrew import Agent, Crew
from langcrew.tools import ToolRegistry

# Create an agent with tools
researcher = Agent(
    role="Research Analyst",
    goal="Find information on any topic",
    backstory="Expert at finding reliable information",
    tools=[
        ToolRegistry.get_tool("crewai:search_the_internet_with_serper"),
        ToolRegistry.get_tool("crewai:read_website_content"),
        ToolRegistry.get_tool("crewai:read_a_files_content")
    ]
)

# Use in a crew
crew = Crew(agents=[researcher])
result = crew.kickoff(inputs={
    "topic": "renewable energy trends"
})

Common Issues

”Tool not found” Error

Step 1: Install the required tool packages

pip install crewai_tools langchain_community

Step 2: Use the correct provider prefix

from langcrew.tools import ToolRegistry

# External tools need provider prefixes (use snake_case)
tool = ToolRegistry.get_tool("crewai:read_a_files_content")  # ✅
tool = ToolRegistry.get_tool("langchain:wikipedia")  # ✅

# Custom tools don't need prefixes
tool = ToolRegistry.get_tool("sales_report")  # ✅

Step 3: Debug with list_tools() if needed

from langcrew.tools import ToolRegistry

# See all available tools (triggers lazy loading)
available_tools = ToolRegistry.list_tools()
print("Available tools:")
for tool in available_tools:
    print(f'  "{tool}"')

# Try accessing a specific tool to see error details
try:
    tool = ToolRegistry.get_tool("crewai:read_website_content")
    print(f"Success: {tool.name}")
except ValueError as e:
    print(f"Error: {e}")

Missing Dependencies Error

Some LangChain tools need additional packages:

# For web search tools
pip install duckduckgo-search

# For Wikipedia tool
pip install wikipedia

# For calculator
pip install numexpr

Creating Custom Tools Not Working

Make sure your tools directory structure is correct:

your_project/
├── tools/
│   ├── __init__.py  # Can be empty
│   └── my_tools.py  # Your tool classes here
└── main.py

Your tool class must:

• Inherit from BaseTool
• Have a name attribute
• Implement the _run method

Pro tip: Start with crewai_tools for common tasks, then create custom tools for your specific business needs.