NogiRuka / GPTCommitVSCode

VS Code extension which helps to generate AI commit messages using ChatGPT

GPT Commit

VS Code extension which helps to generate AI commit messages using ChatGPT.

Features

You can generate commit message by pressing 'Generate AI commit' button in source control tab:

Tip: You could also generate commit from command pallete by calling 'Generate AI commit' command.

Requirements

You need Open AI API Key to make this extension work. You can get your API key from OpenAI.

Extension Settings

GPT Commit extension contributes the following settings:

Appearance

• gptcommit.appearance.delimeter: Delimeter between commit lines

General

• gptcommit.general.generator: Generator used to create commit messages. Available options: ChatGPT
• gptcommit.general.messageApproveMethod: Method used to approve generated commit message. Available options: Quick pick, Message file

OpenAI

• gptcommit.openAI.apiKey: OpenAI API Key. Needed for generating AI commit messages
• gptcommit.openAI.gptVersion: Version of GPT used by OpenAI
• gptcommit.openAI.customEndpoint: Custom endpoint URL for OpenAI API
• gptcommit.openAI.temperature: Controls randomness. Lowering results in less random completions. As the temperature approaches zero, the model will become deterministic and repetitive
• gptcommit.openAI.maxTokens: The maximum number of tokens to generate. Requests can use up to 2048 tokens shared between prompt and completion

Release Notes

1.0.0

Initial release of GPT Commit

1.0.1

Updated icons

1.0.2

Fixed UX

1.0.3

Added Open AI API Key input prompt

1.0.4

Updated commit formatting. Added new setting

1.0.5

Added new advanced configuration for ChatGPT.
Added new option to accept and edit generated commit via temp message file. (Thanks for help chenweiyi)
Added option to set custom ChatGPT endpoint URL. (Thanks for help aiyogg)
Fixed issue with git on windows (Issue #5)
Added option to select different ChatGPT version (Issue #6)
Set default ChatGPT version to gpt-3.5-turbo-16k

License

Released under MIT by @dmytrobaida.