mris.dev

What Is It?

Repo Summarizer is a GitLab repository analyzer that generates business use case summaries, identifies tech stacks, and provides maintenance recommendations. It fetches repository files using the GitLab API and uses a local LLM (via OpenAI-compatible API) to generate comprehensive reports.

Why Build It?

When joining a new project or auditing existing codebases, understanding the tech stack, business purpose, and maintenance state takes significant time. Repo Summarizer automates this analysis, generating detailed reports in minutes instead of hours.

How It Works

Fetches repository files using the GitLab API
Analyzes code structure — identifies frameworks, languages, dependencies
Generates business summary — uses LLM to understand the project’s purpose
Identifies tech stack — detects frameworks, libraries, and tools
Provides recommendations — suggests maintenance improvements

Features

Business use case analysis — Understand what the project does
Tech stack identification — Automatic detection of frameworks and libraries
Maintenance recommendations — Actionable suggestions for improvement
Local LLM support — Works with Ollama or any OpenAI-compatible API
External CLI agents — Can offload to Gemini CLI or similar tools
GitLab integration — Native GitLab API support with PAT authentication

Tech Stack

Layer	Technology
Language	Python 3.8+
LLM	OpenAI-compatible API (Ollama, etc.)
API	GitLab API
Package Manager	uv / pip

Prerequisites

Python 3.8+
Access to the target GitLab instance
GitLab PAT with read_api or read_repository scope
Local LLM (e.g., Ollama) or OpenAI-compatible API

Quick Start

Installation

pip install -r requirements.txt
# OR using uv
uv pip install -r requirements.txt

Configuration

cp .env.example .env

Edit .env:

GITLAB_PAT — Your GitLab Personal Access Token
GITLAB_PROJECT_ID — Project path (e.g., application/service) or numeric ID
LLM_BASE_URL — Local LLM URL (default: http://localhost:11434/v1)

Running with Local LLM

# Start Ollama
ollama run llama3

# Run the analyzer
python main.py

Using External CLI Agents

Can be configured to use external CLI coding agents like gemini-cli instead of local LLMs.

Use Cases

Onboarding — Quickly understand new codebases
Code Audits — Automated repository analysis
Documentation — Generate project summaries automatically
Tech Debt Assessment — Identify maintenance issues
Portfolio Analysis — Analyze multiple repositories for patterns

Example Output

The tool generates reports including:

Business Summary — What the project does and its purpose
Tech Stack — Frameworks, libraries, and tools used
Architecture — High-level structure and patterns
Maintenance Status — Health indicators and recommendations
Dependencies — Key dependencies and their versions

Repo Summarizer