Python
SDK
bashlet
Python SDK for bashlet - provide sandboxed bash execution as tools for AI agents. Supports both sync and async clients.
01 Installation
Terminal
# Core package
pip install bashlet
# With framework support
pip install bashlet[langchain] # LangChain
pip install bashlet[openai] # OpenAI
pip install bashlet[anthropic] # Anthropic
pip install bashlet[mcp] # MCP
pip install bashlet[all] # All frameworksMake sure you have bashlet installed on your system.
02 Quick Start
Synchronous Client
Python
from bashlet import Bashlet, Mount
bashlet = Bashlet(
mounts=[Mount("./src", "/workspace")],
)
# Execute a command
result = bashlet.exec("ls -la /workspace")
print(result.stdout)Asynchronous Client
Python
import asyncio
from bashlet import AsyncBashlet
async def main():
bashlet = AsyncBashlet()
result = await bashlet.exec("echo hello")
print(result.stdout)
asyncio.run(main())03 AI Framework Support
LangChain
BaseTool integration for LangChain agents
OpenAI
Function calling for OpenAI API
Anthropic
Tool use for Claude API
MCP
Model Context Protocol support
LangChain
Python
from langchain_openai import ChatOpenAI
from bashlet import Bashlet
bashlet = Bashlet(
mounts=[{"host_path": "./project", "guest_path": "/workspace"}],
)
# Get LangChain tools
tools = bashlet.to_langchain_tools()
# Bind tools to LLM
llm = ChatOpenAI(model="gpt-4-turbo")
llm_with_tools = llm.bind_tools(tools.all())
# Use in agent
response = llm_with_tools.invoke("List files in /workspace")OpenAI Function Calling
Python
from openai import OpenAI
from bashlet import Bashlet
client = OpenAI()
bashlet = Bashlet()
# Get OpenAI tools
handler = bashlet.to_openai_tools()
# Create completion with tools
response = client.chat.completions.create(
model="gpt-4-turbo",
tools=handler.definitions,
messages=[{"role": "user", "content": "List files"}],
)
# Handle tool calls
for tool_call in response.choices[0].message.tool_calls or []:
result = handler.handle(
tool_call.function.name,
tool_call.function.arguments
)
print(result)Anthropic Tool Use
Python
from anthropic import Anthropic
from bashlet import Bashlet
client = Anthropic()
bashlet = Bashlet()
# Get Anthropic tools
handler = bashlet.to_anthropic_tools()
# Create message with tools
response = client.messages.create(
model="claude-3-opus-20240229",
max_tokens=1024,
tools=handler.definitions,
messages=[{"role": "user", "content": "List files"}],
)
# Handle tool use
for block in response.content:
if block.type == "tool_use":
result = handler.handle(block.name, block.input)
print(result)MCP Server
Python
from mcp.server import Server
from mcp.server.stdio import stdio_server
from bashlet import Bashlet
bashlet = Bashlet()
handler = bashlet.to_mcp_tools()
server = Server("bashlet-server")
@server.list_tools()
async def list_tools():
return handler.definitions
@server.call_tool()
async def call_tool(name: str, arguments: dict):
result = handler.handle(name, arguments)
return result.content
async def main():
async with stdio_server() as (read_stream, write_stream):
await server.run(read_stream, write_stream)04 Available Tools
| Tool | Description |
|---|---|
| bashlet_exec | Execute shell commands in the sandbox |
| bashlet_read_file | Read file contents from the sandbox |
| bashlet_write_file | Write content to a file in the sandbox |
| bashlet_list_dir | List directory contents |
05 API Reference
Bashlet / AsyncBashlet Class
Constructor
Bashlet(
binary_path: str = "bashlet", # Path to bashlet binary
preset: str | None = None, # Default preset name
mounts: list | None = None, # Default mounts
env_vars: list | None = None, # Environment variables
workdir: str | None = None, # Working directory
timeout: int = 300, # Timeout in seconds
config_path: str | None = None, # Config file path
)Methods
| Method | Description |
|---|---|
| exec(command, **options) | Execute a one-shot command |
| create_session(**options) | Create a persistent session |
| run_in_session(id, command) | Run command in existing session |
| terminate(session_id) | Terminate a session |
| list_sessions() | List all active sessions |
| read_file(path) | Read file from sandbox |
| write_file(path, content) | Write file to sandbox |
| list_dir(path) | List directory contents |
| to_langchain_tools() | Generate LangChain tools |
| to_openai_tools() | Generate OpenAI tools |
| to_anthropic_tools() | Generate Anthropic tools |
| to_mcp_tools() | Generate MCP tools |
| to_generic_tools() | Generate framework-agnostic tools |