Metadata-Version: 2.4
Name: bashkit
Version: 0.1.4
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Programming Language :: Rust
Classifier: Topic :: Software Development :: Interpreters
Classifier: Topic :: Security
Requires-Dist: deepagents>=0.3.11 ; extra == 'deepagents'
Requires-Dist: langchain-anthropic>=0.3 ; extra == 'deepagents'
Requires-Dist: pytest>=7.0 ; extra == 'dev'
Requires-Dist: pytest-asyncio>=0.23 ; extra == 'dev'
Requires-Dist: langchain-core>=0.3 ; extra == 'langchain'
Requires-Dist: langchain-anthropic>=0.3 ; extra == 'langchain'
Provides-Extra: deepagents
Provides-Extra: dev
Provides-Extra: langchain
Summary: Python bindings for Bashkit - a sandboxed bash interpreter for AI agents
Keywords: bash,sandbox,ai,agent,shell,interpreter
License: MIT
Requires-Python: >=3.9
Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM

# Bashkit Python Bindings

Python bindings for [Bashkit](https://github.com/everruns/bashkit) - a sandboxed bash interpreter for AI agents.

## Features

- **Sandboxed execution**: All commands run in isolation with a virtual filesystem
- **68+ built-in commands**: echo, cat, grep, sed, awk, jq, curl, find, and more
- **Full bash syntax**: Variables, pipelines, redirects, loops, functions, arrays
- **Resource limits**: Protect against infinite loops and runaway scripts
- **LangChain integration**: Ready-to-use tool for LangChain agents

## Installation

```bash
# From PyPI (when published)
pip install bashkit

# With LangChain support
pip install 'bashkit[langchain]'

# From source
pip install maturin
maturin develop
```

## Quick Start

```python
import asyncio
from bashkit import BashTool

async def main():
    tool = BashTool()

    # Simple command
    result = await tool.execute("echo 'Hello, World!'")
    print(result.stdout)  # Hello, World!

    # Pipeline
    result = await tool.execute("echo -e 'banana\\napple\\ncherry' | sort")
    print(result.stdout)  # apple\nbanana\ncherry

    # Virtual filesystem
    result = await tool.execute("""
        echo 'data' > /tmp/file.txt
        cat /tmp/file.txt
    """)
    print(result.stdout)  # data

asyncio.run(main())
```

## LangChain Integration

```python
from bashkit.langchain import create_bash_tool
from langchain.agents import create_agent

# Create tool
bash_tool = create_bash_tool()

# Create agent
agent = create_agent(
    model="claude-sonnet-4-20250514",
    tools=[bash_tool],
    system_prompt="You are a helpful assistant with bash skills."
)

# Run
result = agent.invoke({
    "messages": [{"role": "user", "content": "Create a file with today's date"}]
})
```

## Configuration

```python
tool = BashTool(
    username="agent",           # Custom username (whoami)
    hostname="sandbox",         # Custom hostname
    max_commands=1000,          # Limit total commands
    max_loop_iterations=10000,  # Limit loop iterations
)
```

## Synchronous API

```python
from bashkit import BashTool

tool = BashTool()
result = tool.execute_sync("echo 'Hello!'")
print(result.stdout)
```

## API Reference

### BashTool

- `execute(commands: str) -> ExecResult`: Execute commands asynchronously
- `execute_sync(commands: str) -> ExecResult`: Execute commands synchronously
- `description() -> str`: Get tool description
- `help() -> str`: Get LLM documentation
- `input_schema() -> str`: Get JSON input schema
- `output_schema() -> str`: Get JSON output schema

### ExecResult

- `stdout: str`: Standard output
- `stderr: str`: Standard error
- `exit_code: int`: Exit code (0 = success)
- `error: Optional[str]`: Error message if execution failed
- `success: bool`: True if exit_code == 0
- `to_dict() -> dict`: Convert to dictionary

## License

MIT

