appleboy/CodeGPT

A CLI written in Go language that writes git commit messages or do a code review brief for you using ChatGPT AI (gpt-4.1, gpt-4o model) and automatically installs a git prepare-commit-msg hook.

CodeGPT – AI‑powered commit‑message and code‑review helper

What it is – A command‑line tool written in Go that talks to large‑language‑model APIs (OpenAI, Azure OpenAI, Gemini, Anthropic, Ollama, Groq, OpenRouter, LiteLLM proxy). It uses the model to:

  • Generate conventional‑commit messages from the current git diff.
  • Produce short code‑review summaries.
  • Translate the generated text into English, Traditional Chinese, Simplified Chinese or Japanese.
  • Stream the model’s output so you see tokens appear in real time.

Why you might care – Writing good commit messages and doing quick code‑review checks are repetitive tasks. CodeGPT automates them by calling a LLM, letting you keep a clean commit history without leaving the terminal.

Key features (as listed in the README)

  • Supports many back‑ends: OpenAI (gpt‑4o, gpt‑4, gpt‑5), Azure OpenAI, Gemini, Anthropic, Ollama, Groq, OpenRouter, LiteLLM proxy.
  • Generates messages that follow the Conventional Commits spec.
  • Installs a prepare‑commit‑msg Git hook automatically.
  • Customizable diff context lines (default 3) and file‑exclusion patterns.
  • Built‑in language translation (en, zh‑tw, zh‑cn, ja).
  • SOCKS or HTTP proxy support.
  • Brief code‑review generation.
  • Streaming output for live token display.
  • Prompt‑template system – you can edit the templates used for commits, reviews, etc.

Installation options

Platform Method
macOS brew tap appleboy/tap && brew install codegpt
Windows choco install codegpt
Linux / any OS Run the provided install script (bash <(curl -sSL https://raw.githubusercontent.com/appleboy/CodeGPT/main/install.sh)) or download a pre‑compiled binary from the releases page and place it in $HOME/.config/codegpt/bin (or any directory in $PATH).
From source go install github.com/appleboy/CodeGPT/cmd/codegpt@latest
VS Code devcontainer Add the feature ghcr.io/kvokka/features/codegpt:1 to devcontainer.json.

The script respects two environment variables:

  • VERSION – pick a specific release.
  • INSECURE=1 – skip SSL verification (useful behind corporate proxies).

Quick start / typical workflow

  1. Configure an API key – e.g. for OpenAI:
    export OPENAI_API_KEY=sk-xxxxxxx   # or store via the CLI
codegpt config set openai.api_key sk-xxxxxxx
    The tool can also fetch keys dynamically with an API‑key helper command (useful with password managers or cloud secret stores).
  2. Install the Git hook (once per repository):
    codegpt hook install   # creates .git/hooks/prepare-commit-msg
  3. Commit as usual – when you run git commit, the hook calls CodeGPT, which:
    • runs git diff on the staged changes,
    • sends the diff to the selected LLM,
    • receives a Conventional‑Commit formatted message and inserts it into the commit editor.
  4. Optional code‑review – run:
    codegpt review
    to get a short summary of the changes.

Configuration highlights

  • Provider selection – openai.provider can be set to openai, azure, gemini, anthropic, litellm, etc.
  • Model choice – openai.model defaults to gpt‑4o but any model supported by the chosen provider can be specified.
  • Proxy – openai.proxy (HTTP/HTTPS) or openai.socks for SOCKS5.
  • Prompt folder – defaults to $HOME/.config/codegpt/prompt; change with codegpt config set prompt.folder /my/prompt.
  • Dynamic key helper – set openai.api_key_helper to a shell command that returns the key; the result is cached for 15 min (configurable).
  • Output language – output.lang (en, zh‑tw, zh‑cn, ja).

Example commands

# Generate a commit message (CLI mode, no hook)
codegpt commit

# Review the staged changes
codegpt review

# Show the version
codegpt version

Who might use it?

  • Developers who already use Git and want AI‑assisted commit messages without leaving their terminal.
  • Teams that enforce Conventional Commits and need a quick way to keep the format consistent.
  • Anyone who wants a lightweight, configurable LLM client for on‑the‑fly code summarisation.

License & community

The repository includes CI badges for linting, testing, security scanning and code‑coverage, indicating an actively maintained open‑source project. The README links to a Chinese blog post and a YouTube demo for additional context.

Bottom line – CodeGPT is a practical, Go‑based CLI that plugs LLMs into the Git workflow, automating commit‑message generation and brief code reviews while offering extensive configuration (multiple providers, proxy support, dynamic credential helpers, custom prompts).

Related

  • Project
    tak-bro/aicommit2

    A reactive CLI tool that automatically generates commit messages for Git, YADM, and Jujutsu using various AI providers.

  • Project
    Nutlope/aicommits

    A CLI tool that uses AI to analyze staged git changes and automatically generate descriptive commit messages in various formats.

  • Project
    guanguans/ai-commit

    A CLI tool that uses AI to automatically generate standardized conventional git commit messages based on code changes.

  • Project
    Significant-Gravitas/AutoGPT

    AutoGPT is an open‑source platform for building, deploying, and managing autonomous AI agents. Users can describe a goal in plain English or assemble a workflow visually; the system creates an LLM‑driven agent that interacts with 45+ integrations (Gmail, Slack, GitHub, etc.) to complete tasks such as drafting reports, triaging incidents, or generating marketing copy. The project offers a paid hosted service and a free self‑hostable Docker appliance, with the core platform under a Polyform Shield license and the classic agent code under MIT.

  • Project
    zurawiki/gptcommit

    A git prepare-commit-msg hook that uses OpenAI language models to automatically generate clear and descriptive commit messages based on staged changes.