Built upon the code from of Aider[https://github.com/paul-gauthier/aider] / Auto-GPT[https://github.com/Significant-Gravitas/Auto-GPT] and GPT4, r2 is for the new era of prompt-based developers.

R2 can be prompted to modify or create new code, it can develop it's own test's, execute the tests and debug/repair mistakes as they arise. When the changes are testing successfully, r2 will prompt you to commit to the repo, complete with message.

With the tool, you can ask for features, improvements, or bug fixes and r2 will apply the suggested changes to source files.

Note: this version has been a self-referencing tool, used to develop itself. My intention is to add ReAct based logic, method by method editing and a jupyter style GUI. Sending an entire file to GPT-4 is expensive and quickly limited by context size.

To use r2, you will require your own GPT-4 API key. 'r2' will prompt to execute .py files through unit tests and commands. This could result in bad things happening to the operating system and it's files if not carefully monitored.

Happy to chat, available on - Discord[https://discord.gg/YaqMe653] or Twitter.

• Chat with GPT-4 about your code by launching r2 from the command line with set of source files to discuss and edit together.
• Request new features, changes, improvements, or bug fixes to your code. Ask for new test cases, updated documentation or code refactors.
• r2 will apply the edits suggested by GPT-4 directly to your source files.
• r2 will automatically commit each changeset to your local git repo with a descriptive commit message. These frequent, automatic commits provide a safety net. It's easy to undo r2 changes or use standard git workflows to manage longer sequences of changes.
• r2 can review multiple source files at once and make coordinated code changes across all of them in a single changeset/commit.
• r2 knows about all the files in your repo, so it can ask for permission to review whichever files seem relevant to your requests.
• You can also edit the files in your editor while chatting with r2.
  • r2 will notice if you edit the files outside the chat.
  • It will help you commit these out-of-band changes, if you'd like.
  • It will bring the updated file contents into the chat.
  • You can bounce back and forth between the r2 chat and your editor, to fluidly collaborate.
• Live, colorized, human friendly output.
• Readline style chat input history, with autocompletion of code tokens found in the source files being discussed (via prompt_toolkit and pygments lexers)

1. Install the package: pip install git+https://github.com/InDate/r2.git
2. Set up your OpenAI API key as an environment variable OPENAI_API_KEY or by including it in a .env file.

Run the r2 tool by executing the following command:

r2

You can launch r2 anywhere in a git repo without naming files on the command line. It will discover all the files in the repo. You can then add and remove individual files in the chat session with the /add and /drop chat commands described below.

You can also use additional command-line options to customize the behavior of the tool. The following options are available, along with their corresponding environment variable overrides:

• --input-history-file INPUT_HISTORY_FILE: Specify the chat input history file (default: .r2.input.history). Override the default with the environment variable r2_INPUT_HISTORY_FILE.
• --chat-history-file CHAT_HISTORY_FILE: Specify the chat history file (default: .r2.chat.history.md). Override the default with the environment variable r2_CHAT_HISTORY_FILE.
• --no-pretty: Disable pretty, colorized output. Override the default with the environment variable r2_PRETTY (default: 1 for enabled, 0 for disabled).
• --no-auto-commits: Disable auto commit of changes. Override the default with the environment variable r2_AUTO_COMMITS (default: 1 for enabled, 0 for disabled).
• --show-diffs: Show diffs when committing changes (default: False). Override the default with the environment variable r2_SHOW_DIFFS (default: 0 for False, 1 for True).
• --yes: Always say yes to every confirmation (default: False).

For more information, run r2 --help.

r2 supports the following commands from within the chat:

• /add <file>: Add matching files to the chat session.
• /drop <file>: Remove matching files from the chat session.
• /ls: List all known files and those included in the chat session.
• /commit [message]: Commit outstanding changes to the chat session files. Use this to commit edits you made outside the chat, with your editor or git commands.
• /undo: Undo the last git commit if it was done by r2.
• /diff: Display the diff of the last r2 commit.
• /debug <file>: Provide a program that is failing and r2 will execute file, upload relevant files and errors and then offer suggestions on how to fix.
• /unit_test <file>: Used to unit test a program file. r2 will prompt to build unit_tests if not available.
• /spec_file <file>: Build a code2text specification. Useful for describing the whole program in small context windows
• /test_connection: Test the connection to GPT without sending additional files or chat context
• /execute_file <file>: Used to execute files and commands.

To use a command, simply type it in the chat input followed by any required arguments.

• Large changes are best performed as a sequence of bite sized steps. Same as if you were undertaking them by yourself.
• Use Control-C to safely interrupt r2 if it isn't providing a useful response. The partial response remains in the conversation, so you can refer to it when you reply with more information or direction.
• Enter a multiline chat message by entering { alone on the first line. End the multiline message with } alone on the last line.
• If your code is throwing an error, paste the error message and stack trace into r2 as a multiline {} message and let r2 fix the bug.
• GPT-4 knows about a lot of standard tools and libraries, but may get some of the fine details wrong about APIs and function arguments. You can paste doc snippets into the chat with the multiline {} syntax.
• r2 will notice if you launch it on a git repo with uncommitted changes and offer to commit them before proceeding.
• r2 can only see the content of the files you specify, but it also gets a list of all the files in the repo. It may ask to see additional files if it feels that's needed for your requests.

You can only use r2 to edit code that fits in the GPT context window. For GPT-4 that is 8k tokens. It helps to be selective about how many source files you discuss with r2 at one time. You might consider refactoring your code into more, smaller files (which is usually a good idea anyway). You can use r2 to help perform such refactorings, if you start before the files get too large.