BATCH_01 // #04

GitScripe

My Experiment with AI Agents

GitScripe was born from my curiosity about LLM agents. I wanted to see if I could build a system that doesn't just list commits, but actually 'understands' the intent behind them. I learned how to orchestrate multiple agents (Analyzer, Summarizer, Critic) and implemented a RAG pipeline so I could chat with my own codebase history.

Project Highlights

[+]Built a multi-agent pipeline to process and critique Git diffs for better summaries.
[+]Implemented a RAG (Retrieval-Augmented Generation) system for repository-aware chat.
[+]Learned background job orchestration with Redis and BullMQ for processing large repos.
[+]Developed a GitHub-style dark UI to present AI insights in a familiar developer context.

Tech Stack

#Node.js#React#TypeScript#Tailwind CSS#Prisma#Redis#Ollama

Open Source Repository

Challenge

Commit messages are often vague. I wanted to use AI to automatically infer intent and summarize changes accurately.

Solution

A multi-stage agentic pipeline that analyzes diffs and groups them by architectural impact.

Discover Repositories

Available Repositories (4)

ajay99511/Social_Networking_Application

Synced

Full-stack social platform built with .NET and Angular.

ajay99511/GitScripe

Synced

Agentic Git intelligence platform using RAG and LLMs.

ajay99511/DL_Algorithms

From-scratch implementations of core Deep Learning models.

ajay99511/RepoPulse

Personal workspace management dashboard.