Skip to main content
The Local SDK provides a set of methods for managing files in your project directory. This guide explains how to use these methods, referencing the actual implementation in our LangGraph React App. At the end, you’ll find a unified example script that demonstrates all file operations together.

​
Overview of File Tools

The main file management methods are:
  • create_file(file_path, content): Create or overwrite a file with the given content.
  • read_file(file_path): Read and return the content of a file.
  • search_files(query, file_patterns): Search for lines matching a query across files.
All file paths are relative to your SDK’s working directory. The SDK will automatically create parent directories if they don’t exist.
Tip: The SDK is designed for text files. For binary files, use encoding/decoding (see below).

​
Method Reference

​
1. Creating Files

def create_file(file_path: str, content: str) -> bool
  • file_path (str): Path to the file, relative to working_dir
  • content (str): Text content to write
  • Returns (bool): True on success, False otherwise

​
2. Reading Files

def read_file(file_path: str) -> Optional[str]
  • file_path (str): Path to the file, relative to working_dir
  • Returns (str or None): File content as string, or None if the file does not exist

​
3. Searching Files

def search_files(query: str, file_patterns: Optional[List[str]] = None) -> Dict[str, List[str]]
  • query (str): Substring or regex to match
  • file_patterns (List[str], optional): Glob patterns for files to search (default: ['**/*'])
  • Returns (Dict[str, List[str]]): Mapping of file paths to lists of matching lines

​
Unified Example: All File Operations

Below is a complete script that demonstrates how to use all file management commands together. This example is based on the actual usage in react.py: 
from cloudcode import Local
import os

# Set your project directory and API key
your_project_dir = "/path/to/your/project"
sdk = Local(
    working_dir=your_project_dir,
    api_key=os.getenv("CLOUD_CODE_API_KEY"),
)

# 1. Create a new file
file_path = "src/example.py"

file_content = """
def hello(name):
    return f"Hello, {name}!"
"""

created = sdk.create_file(file_path, file_content)
if created:
    print(f"Created file: {file_path}")
else:
    print(f"Failed to create file: {file_path}")


# 2. Read the file back
content = sdk.read_file(file_path)
if content is not None:
    print(f"\nRead from {file_path}:\n{content}")
else:
    print(f"File {file_path} does not exist or could not be read.")


# 3. Search for a function definition in all Python files
search_query = "def hello"
search_results = sdk.search_files(
    query=search_query,
    file_patterns=["src/**/*.py"]
)
print(f"\nSearch results for '{search_query}':")
for path, lines in search_results.items():
    print(f"- {path}")
    for line in lines:
        print(f"    {line.strip()}")


# 4. (Optional) Handling binary files
import base64
binary_data = b'\x00\x01\x02\x03'
encoded = base64.b64encode(binary_data).decode('utf-8')
sdk.create_file("data/binary.dat", encoded)
read_encoded = sdk.read_file("data/binary.dat")
if read_encoded:
    binary_content = base64.b64decode(read_encoded)
    print(f"\nRead {len(binary_content)} bytes from binary.dat")

​
Best Practices

  • Use relative paths for portability.
  • Check file existence before reading or overwriting.
  • Use specific glob patterns in search_files for performance.
  • Process search results programmatically to identify relevant files.
  • Handle binary files with encoding/decoding as shown above.

​
More Advanced Usage

You can combine file operations with the code() and code_headless() methods for AI-powered code editing. See the AI Coding page for details.