This command-line GIT copilot analyzes your last commit and leverages an LLM of your choice to enhance it.
Project description
Cocommit: A Copilot for Git
Cocommit is a command-line tool that works with your HEAD commit and leverages an LLM of your choice to enhance commit quality.
A good commit consists of multiple elements, but at a minimum, it should have a well-crafted commit message. Cocommit analyzes the message from the last (HEAD) commit and suggests improvements, highlighting both strengths and areas for enhancement.
Cocommit v2 is currently in development and will introduce many new features—see the v2 documentation for details.
Cocommit utilizes LangChain as an abstraction layer to access various Large Language Models (LLMs).
Why Use Cocommit?
- Works with the LLM of your choice
- Provides a simple foundation for building custom AI-powered tools
- Easily integrates into your development workflow
- Allows customization of the LLM experience to fit your project needs
Table of Contents
Example CLI Session
Session Start
git add .
git commit
cocommit -s bedrock-claude37
Execution Output
Calling with: --model_provider bedrock --model us.anthropic.claude-3-7-sonnet-20250219-v1:0 --region_name us-east-1
Calling LLM....
Done in 9.6 seconds.
About your commit:
This is a good quality commit message that follows most best practices. It has a clear, concise title in imperative mood and a brief explanation of why the change was made.
Strengths:
- Concise first line under 80 characters
- Uses imperative mood correctly ("Add" not "Added")
- Provides context in the description about why the change was made
- Clearly specifies what was changed
- Follows proper structure with title, blank line, and description
Improvements:
- The description could be slightly more specific about which Python versions are now supported
********** Proposed Git Message: ********************
Add Python version classifiers to project metadata
This change lists supported Python versions for better visibility on PyPI.
****************************************************
Fixes:
- Consider specifying which Python versions are now supported in the description for more detail
Amend the commit message? [Y/n]: y
********** Previous Message **********
Add Python version classifiers metadata
Lists supported Python versions for PyPI.
**************************************
Amend ... done!
Installation
To install Cocommit, run:
pip install cocommit
Installing from Source
You can also install Cocommit from the source code, which allows for customization. For details, see the Contributing Guide.
Handling on-demand Dependencies
When calling a specific LLM, you may encounter a message like this:
cocommit --model llama3-8b-8192 --model_provider groq
Unable to import langchain_groq. Please install with `pip install -U langchain-groq`
To resolve this, execute the suggested pip install command. Since Cocommit supports a wide variety of LLMs, dependencies are installed only when needed.
Usage
Cocommit interacts with Large Language Models (LLMs) through an abstraction layer provided by LangChain. This allows you to use any LLM provider supported by LangChain’s init_chat_model function.
Example: Using OpenAI
To use OpenAI, set your API key as the OPENAI_API_KEY environment variable. Alternatively, you can provide the key via the command line using the --api_key <your key> option:
cocommit --model_provider openai --model gpt-4o
Note: On the first run, you may need to install additional dependencies by running: pip install -U langchain-openai.
Simplifying OpenAI Usage with Shortcuts
Cocommit provides shortcuts for common LLM providers and models, allowing you to avoid specifying every parameter manually.
Ensure your chosen LLM provider is authorized (e.g., OpenAI API key or AWS credentials).
Run:
cocommit -s NAME
For example, if OPENAI_API_KEY is set:
cocommit -s gpt-4o
This command is equivalent to the full OpenAI GPT-4o example above.
Example: Using Claude 3.7 on Bedrock
Before using Amazon Bedrock, ensure that:
- The Claude 3.7 model is enabled in your AWS account (region:
us-east-1) - You have valid credentials to access the model (typically configured in
~/.aws/credentialsor via environment variables; refer to AWS documentation for details)
To analyze and enhance the last commit message in a Git repository, run:
cocommit --model_provider bedrock --model us.anthropic.claude-3-7-sonnet-20250219-v1:0 --region us-east-1
To simplify execution for future use, save the above command in a shell script.
Simplifying Bedrock Claude 3.7 Usage with Shortcuts
If AWS credentials are configured and Claude 3.7 is enabled in us-east-1, run:
cocommit -s bedrock-claude37
This command is equivalent to the previous Bedrock example.
Viewing Available Shortcuts
To see all available shortcuts, run:
cocommit --show-shortcuts
If a shortcut you need is missing, consider contributing! See the Contributing Guide for details.
Using Other LLMs
Cocommit leverages LangChain’s init_chat_model function to call different LLMs. Command-line arguments map directly to its parameters.
- The
--modelparameter corresponds to themodelargument ininit_chat_model. It's the first positional parameter in LangChain's documentation. - Other command-line parameters are passed directly to
init_chat_model(ensure you use_instead of-when required by LangChain’s documentation).
Recommended usage: Always specify --model and --model_provider explicitly, unless using a shortcut.
- init_chat_model documentation
- models (certain examples demonstrate the use of init_chat_model with its required parameters)
Useful CLI Options
Cocommit supports debugging features, including:
- Viewing the raw LLM prompt
- Displaying the raw LLM response
To explore available options, run:
cocommit --help
Contributing
Contributions are appreciated! If you'd like to get started, please review the contributing guidelines.
License
This project is licensed under the MIT License.
