rdmgen

rdmgen is a Python CLI tool to automatically generate high-quality README.md files for code repositories using Amazon Bedrock and LangGraph.

Overview

This project uses the Amazon Bedrock, LangGraph, and LangChain AWS libraries to analyze the structure and contents of a code repository and generate a comprehensive README.md file. The generated README provides an overview of the project, installation instructions, usage examples, and other relevant details based on the files in the repository.

Workflow

Features

• Automated README Generation: No more manually writing README files! rdmgen does the hard work for you.
• Local or Remote Repositories: Supports generating README files for local codebases or remote Git repositories.
• Execution Visualization: Optionally generate a diagram visualizing the LangGraph agent's execution flow.
• Customizable Output: Write the generated README.md file to any directory you specify.

Installation

Please Note: You need python >= 3.9. You can check this with python --version. If your base environment does not have the required version, create a virtual environment with the required version and use "Option 1" to install.

Option 1: Using pip

You can install the package with pip after cloning the repository.

pip install .

Option 2: Makefile

A Makefile with some recipes is available for linux/MacOS. To create a dedicated Python venv and install rdmgen in the virtual environment, run:

make all

The virtual env is named .venv by default, and it's located in the same directory as the repository.

Now run the following command to activate the environment:

source .venv/bin/activate

For more make recipes such as cleaning up and removing the venv, check the output of make help.

Usage

Prerequisites

1. AWS credentials: Before running the cli commands to generate a README file, AWS credentials are required. If you are running this from within an AWS environment (e.g. SageMaker Notebook instance), you can proceed to step 2 to ensure correct permissions are granted to the attached IAM Role.

2. IAM Permissions: The IAM User or Role should have at least the following permissions on the selected model (replace the resource * with the model id(s)):

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "VisualEditor",
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel",
                "bedrock:InvokeModelWithResponseStream"
            ],
            "Resource": "*"
        }
    ]
}

3. Model access in Amazon Bedrock: ensure you have access to the selected model via the AWS Console under the Amazon Bedrock Service.

Note: The project uses Claude 3 sonnet: anthropic.claude-3-sonnet-20240229-v1:0 from Amazon Bedrock. Please make sure you have access to it. If you would like to change the model ID, please update it in the src/rdmgen/constants.py and re-install the package.

Commands

To generate a README.md file for your repository, run the cli command:

rdmgen <REPO_PATH> --out <OUTPUT_FOLDER>

• <REPO_PATH>: can be a local clone/directory or a remote Git URL.
• <OUTPUT_FOLDER>: path to save the output README.md file to. If not specified, the file is saved to the same directory of the source code.

You can pass the optional --diagram flag to create the flowchart of the agent to see the steps it ran.

Examples

The following two examples are availabe in this repository for testing:

rdmgen test_data/hello_world

rdmgen test_data/sensor_extraction --out ./output

Example for a small remote repository on GitHub:

rdmgen https://github.com/pypa/sampleproject --out output/sampleproject --diagram

Limitations

The rdmgen utility works currently for small to mid-size repositories, consider the following adjustable limitations for large repositories, which are defined in src/constants.py:

• Amazon Bedrock read-timeout, which is passsed to the Bedrock runtime client as a config, more details: https://repost.aws/knowledge-center/bedrock-large-model-read-timeouts
• LangGraph recursion_limit, which is passed to the configuration when invoking the graph. More details: https://langchain-ai.github.io/langgraph/how-tos/recursion-limit/?h=recursion

Development

For development and extensions, this project uses Poetry for dependency management. To install the package and the dependencies, install poetry and then run:

poetry lock 
poetry install

Security

See CONTRIBUTING for more information.

Contributing

Contributions are welcome! Please open an issue or submit a pull request if you have any improvements or bug fixes.

Potential Improvements:

• Add a Tool to decide which files to read, i.e. make it optional to read the un-important files.
• Read the files iteratively for each README section. i.e. Firstly create a README template, then further loop through each section and read only the files that might contribute content to that particular section. Perhaps a pattern like: https://langchain-ai.github.io/langgraph/tutorials/plan-and-execute/plan-and-execute/

License

This library is licensed under the MIT-0 License. See the LICENSE file.