Semantic Codebase Search

A simple tool for semantic code search using gpt-oss and embeddings.

Tech Stack

• Python (core logic)
• Go (optional CLI)
• Ollama (local LLM inference)
• FAISS (vector search)
• LangChain (embeddings)

Embedding Model

• all-minilm:33b

Reasoning AI Model

• gpt-oss:20b

Setup Instructions

1. Clone the repository

git clone https://github.com/harryfrzz/semantic-codebase-search.git
cd semantic-codebase-search

2. Install Go (for CLI, optional)

# On Ubuntu/Debian
sudo apt update
sudo apt install -y golang-go

# Check Go version
go version

3. Install Python dependencies

# Make sure you have Python 3.11+
# Install all Python dependencies
pip install -r core_main/requirements.txt

4. Install Ollama and Pull Models

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Start Ollama (in background)
ollama serve &

# Pull required models
ollama pull all-minilm:33m
ollama pull gpt-oss:20b

Usage

Python Script

python3 core_main/entry_point.py "your search query here"

Go CLI (optional)

# Build the CLI
go build -o semantic-search main_cli.go

# Run the CLI
./semantic-search

# Or
go run main_cli.go