Kaguya is a ChatGPT plugin enabling users to interact with local files and execute Python, JavaScript, and bash scripts directly from ChatGPT. It targets developers seeking to integrate their development environment with AI-powered code generation and manipulation, offering a controlled way to manage local codebases.

How It Works

Kaguya operates as a local server, accessible via Docker, that exposes a REST API for file system operations and command execution. It restricts file access to a designated FILES directory, ensuring a controlled environment. The plugin leverages ChatGPT's conversational interface to interpret user commands, translate them into API calls, and execute them on the local machine.

Quick Start & Requirements

• Install Docker and run docker.sh.
• Clone the repository.
• Ensure localhost port 3000 is available.
• Access to OpenAI's plugin devtools is required.
• Files for interaction should be placed in the FILES directory.
• Official Docs: https://github.com/ykdojo/kaguya
• Demo Videos: https://github.com/ykdojo/kaguya/assets/107422421/c580a6f6-5f08-43fd-ac8b-c12a319e1534, https://github.com/ykdojo/kaguya/assets/107422421/d61b8ff1-2dbd-4eb4-b1b5-45d43797ddaa

Highlighted Details

• Supports file operations: list, read, update (search/replace, whole file), append, create, delete, rename.
• Enables execution of Python, JavaScript, and bash scripts.
• API endpoints are documented in openapi.yaml.
• Recommended custom instructions for effective ChatGPT interaction are provided.

Maintenance & Community

• Discord server available for community support.
• VS Code extension waitlist is open.

Licensing & Compatibility

• License details are not explicitly stated in the README.

Limitations & Caveats

File operations are restricted to the FILES directory. Large files (over ~600 lines for reading, ~80 lines for writing) may cause performance issues or task failures. The plugin relies on OpenAI's plugin devtools, which may have a waitlist.