gptify is a command-line tool that transforms a Git repository into a single text file or multiple text chunks suitable for use with Large Language Models (LLMs) like ChatGPT. It preserves the file structure and content, enabling LLMs to understand and process the codebase for tasks such as code review, documentation generation, and answering questions about the code. This project is a fork of gptrepo with added features.
This tool addresses the challenge of effectively using LLMs with codebases. By converting a repository into a digestible format, gptify allows developers to leverage the power of LLMs for various development tasks. It simplifies the process of feeding code context into LLMs, avoiding size limitations and formatting issues.
The easiest way to install gptify is using pip:
pip install gptifyAlternatively, you can install it using pipx:
poetry build && pipx install dist/*.whlYou can also uninstall older versions using the provided install script:
./install.sh- Navigate to the root directory of your Git repository.
- Run the
gptifycommand:
gptifyThis will generate a file named gptify_output.txt in the current directory containing the formatted repository content. You can then copy and paste the contents of this file into a ChatGPT session.
--output <filename>: Specifies the name of the output file (default:gptify_output.txt).--clipboard: Copies the output directly to the clipboard instead of creating an output file.--openfile: Opens the output file after creation using the default system application.--preamble <filepath>: Prepends a custom preamble to the output file. This is useful for providing instructions or context to the LLM.--chunk: Enables chunking of the output into smaller files, useful for handling large repositories that exceed LLM context limits. Used with--max_tokensand--overlap.--max_tokens: Sets the maximum number of tokens per chunk when using the--chunkoption (default: 900000). Requires thetiktokenlibrary.--overlap: Sets the number of overlapping tokens between chunks when using the--chunkoption (default: 400). Helps maintain context across chunks. Requires thetiktokenlibrary.--output_dir: Specifies the output directory for chunks when using--chunk(default:gptify_output_chunks).
gptify --output my_repo.txt --preamble instructions.txtThis command will generate my_repo.txt with the processed repository data, prepended with the content of instructions.txt.
gptify --chunk --max_tokens 4000 --overlap 200This will create multiple files in the gptify_output_chunks directory, each containing a chunk of the repository data, with a maximum of 4000 tokens and an overlap of 200 tokens.
Contributions are welcome.
This project is licensed under the MIT License - see the LICENSE file for details.