Meeting Notes

Meeting Summary

The prototype can be a CLI simulation/documentation showing commands being typed and the output printed.

Primary Goal: The tool must summarize the code differences (the diff) between commits/versions, explaining what changed without requiring the user to read the raw code diff or switching between versions.
Short-Term: The initial focus is on ensuring the tool works with local Git repositories.
Integration with GitHub or GitLab APIs is a low-priority future goal.

LLM Provider: The LLM is external and must be configurable (currently the LLM should be Gemini). The team must include an option for the user to provide their own API key/endpoint.
Input/Output: The tool interacts with a local Git repository folder, and output is a plain text to the CLI with the option to export to a file.

Core Summary: Must provide a summary of code differences (not commit messages).
LLM Configuration: Must allow the user to input the LLM API key and endpoint URL.
Contributor Analysis: Future feature to identify the person responsible for a specific change and summarize their other contributions.
Command Structure: Must include a basic help command.

Need for a user warning if the diff data being sent to the LLM is excessive (e.g., over 50,000 lines/tokens), requiring user confirmation due to potential API costs.
Strong customer preference for the tool to be written in Python.
We should use modern Python dependency management like pyproject.toml.

LLM Dependency: The project is dependent on the customer providing their own Gemini LLM access/token.
Scope: Focus is limited to local Git interaction for now.

An architecture that focuses on four main layers:

Command Layer: Handles user input via the CLI (e.g., help, summarize).
Git Interaction / Parser Layer: Extracts the necessary diffs and contributor metadata from the local repository.
LLM Service Layer: Manages API configuration, context size warnings, and makes the POST request to the external LLM provider.
Output Layer: Prints the final, summarized text directly to the CLI console.

General Meeting Outcome	Actionable Item (WHO, WHAT, WHEN)
We need to figure out how to demonstrate the CLI tool	Team Member to Research and select a Python library suitable for creating the CLI simulation prototype by The end of sprint-2.
We need a parsing component ready before we can summarize.	Team Members to Implement the Git parsing layer to extract and format a simple commit diff range from a local repo by End of sprint-3.
We should have the ability to identify contributions by author.	Team Members to Draft mock-up text and logic for the Contributor Analysis feature, showing how user input would trigger the contributor's summary, by End of sprint 4.