Tool Class

The Tool class represents a generated tool and provides methods for execution and framework adaptation.

Properties

Property	Type	Description
name	str	Tool identifier.
intent	str	Natural language description.
inputs	list[InputParam]	Input parameter definitions.

Execution Methods

run()

Execute the tool and return the result.

result = tool.run(**kwargs) -> Any

Parameters

Accepts keyword arguments matching the tool’s input parameters.

Returns

The result from the tool’s run() function.

Raises

RuntimeError - If execution fails and self-healing is disabled or exhausted.

Example

result = tool.run(query="python tutorials", limit=10)
print(result)

---

run_safe()

Execute the tool without raising exceptions.

result = tool.run_safe(**kwargs) -> ToolResult

Returns

ToolResult - Object with success, data, and error fields.

Example

result = tool.run_safe(query="test")

if result.success:
    print(result.data)
else:
    print(f"Error: {result.error}")

---

run_async()

Execute the tool asynchronously.

result = await tool.run_async(**kwargs) -> Any

Example

import asyncio

async def main():
    result = await tool.run_async(query="test")
    print(result)

asyncio.run(main())

---

run_safe_async()

Execute the tool asynchronously without raising exceptions.

result = await tool.run_safe_async(**kwargs) -> ToolResult

Example

async def main():
    result = await tool.run_safe_async(query="test")
    if result.success:
        print(result.data)

---

run_interactive()

Collect inputs via CLI prompts and execute.

result = tool.run_interactive() -> Any

Prompts the user for each input parameter, then executes the tool.

Example

# Prompts user for inputs
result = tool.run_interactive()
# Enter query: python tutorials
# Enter limit [10]:
# Result: {...}

Framework Adapters

to_langchain()

Convert to a LangChain BaseTool.

lc_tool = tool.to_langchain() -> BaseTool

Returns

BaseTool - LangChain-compatible tool.

Requirements

pip install "anvil-agent[langchain]"

Example

from langchain.agents import create_tool_calling_agent

lc_tool = tool.to_langchain()
agent = create_tool_calling_agent(llm, [lc_tool], prompt)

---

to_crewai()

Convert to a CrewAI BaseTool.

crew_tool = tool.to_crewai() -> BaseTool

Returns

BaseTool - CrewAI-compatible tool.

Requirements

pip install "anvil-agent[crewai]"

Example

from crewai import Agent

crew_tool = tool.to_crewai()
agent = Agent(role="Researcher", tools=[crew_tool])

---

to_autogen()

Convert to an AutoGen FunctionTool.

ag_tool = tool.to_autogen() -> FunctionTool

Returns

FunctionTool - AutoGen-compatible tool.

Requirements

pip install "anvil-agent[autogen]"

Example

from autogen_agentchat.agents import AssistantAgent

ag_tool = tool.to_autogen()
agent = AssistantAgent(name="assistant", tools=[ag_tool])

---

to_openai_agents()

Convert to an OpenAI Agents SDK FunctionTool.

oai_tool = tool.to_openai_agents() -> FunctionTool

Returns

FunctionTool - OpenAI Agents-compatible tool.

Requirements

pip install "anvil-agent[openai-agents]"

Example

from agents import Agent

oai_tool = tool.to_openai_agents()
agent = Agent(name="assistant", tools=[oai_tool])

Chaining

pipe()

Chain this tool with another tool.

chain = tool.pipe(other: Tool) -> ToolChain

Parameters

Parameter	Type	Description
other	Tool	Tool to chain after this one.

Returns

ToolChain - A chain of tools for sequential execution.

Example

chain = search_tool.pipe(summarize_tool).pipe(translate_tool)
result = chain.run(query="AI news")

Information Methods

get_inputs()

Get the list of input parameters.

inputs = tool.get_inputs() -> list[InputParam]

Returns

list[InputParam] - List of input parameter definitions.

Example

for param in tool.get_inputs():
    print(f"{param.name}: {param.param_type} (required={param.required})")

---

has_required_inputs()

Check if the tool has any required inputs.

has_required = tool.has_required_inputs() -> bool

Returns

bool - True if any input is required.

ToolResult

Returned by run_safe() methods.

from anvil.models import ToolResult

result = ToolResult(
    success: bool,
    data: Any = None,
    error: str | None = None,
)

Fields

Field	Type	Description
success	bool	Whether execution succeeded.
data	Any	Result data on success.
error	str | None	Error message on failure.

Example

result = tool.run_safe(query="test")

if result.success:
    process(result.data)
else:
    log_error(result.error)

ToolChain

Created by pipe() method.

Methods

Method	Description
run(**kwargs)	Execute chain, raising on error.
run_safe(**kwargs)	Execute chain safely.
run_async(**kwargs)	Execute chain asynchronously.
run_safe_async(**kwargs)	Execute chain async safely.
pipe(tool)	Add another tool to the chain.

Properties

Property	Type	Description
len(chain)	int	Number of tools in chain.
str(chain)	str	String representation.

Example

from anvil import ToolChain

# Create chain
chain = ToolChain([tool1, tool2, tool3])

# Or use pipe
chain = tool1.pipe(tool2).pipe(tool3)

# Execute
result = chain.run(initial="data")
print(len(chain))  # 3
print(chain)  # ToolChain(tool1 -> tool2 -> tool3)

See Also

• Anvil Class Reference
• Tool Chaining
• Framework Adapters