Open Source Compass

Open Source Compass is an AI-powered platform that helps developers discover and contribute to open source projects that match their skills and interests. It leverages semantic search, vector databases, and Google Gemini LLM via Vertex AI to provide actionable contribution guidelines for selected issues.

---

🚀 Features

• Ingests top issues and docs from open source repos into MongoDB
• Creates vector embeddings for semantic search
• Matches user queries/skills to relevant issues
• Streams AI-generated contribution guidelines using Gemini LLM
• Interactive Streamlit web app for search and exploration

---

🏗️ Architecture

The architecture diagram below illustrates the workflow of Open Source Compass. It starts with fetching issues and documentation from GitLab, which are processed by an ingestion script to generate vector embeddings stored in MongoDB. The Streamlit web app enables users to search for relevant issues using semantic search, and selected issues are sent to Google Gemini LLM via Vertex AI to generate actionable contribution guidelines.

graph TD
    A[GitLab API] -->|Issues & Docs| B[Ingestion Script<br>copyrepotomongo.py]
    B -->|Embeddings| C[MongoDB<br>Vector Store]
    C -->|Vector Search| D[Streamlit Web App<br>app.py]
    D -->|User Query/Skills| C
    D -->|Selected Issue + Docs| E[Gemini LLM<br>Vertex AI]
    E -->|Contribution Guidelines| D

Loading

---

📝 How It Works

1. Ingestion:

   • copyrepotomongo.py fetches issues and docs from GitLab, generates embeddings, and stores them in MongoDB with vector indexes.

2. Semantic Search:

   • Users search by text or skills in the Streamlit app.
   • The app queries MongoDB using vector search to find relevant issues.

3. Guideline Generation:

   • When a user selects an issue, the app fetches the issue and related docs.
   • Sends them to Gemini LLM (Vertex AI) to generate actionable contribution guidelines, streamed back to the UI.

---

🛠️ Setup

1. Clone the repo & install dependencies

   git clone <repo-url>
   cd google-ai-in-action-hackathon
   python -m venv .venv
   source .venv/bin/activate  # or .venv\Scripts\activate on Windows
   pip install -r requirements.txt

2. Configure environment

   • Fill out .env with your MongoDB URI, GitLab token, and Google Cloud project info.
   • Authenticate with Google Cloud:

     gcloud auth application-default login

3. Ingest data

   python Open\ Source\ Compass/copyrepotomongo.py

4. Run the app

   streamlit run Open\ Source\ Compass/app.py

---

📊 Example Workflow

1. Search for issues by text or skills.
2. Review matching issues and select one.
3. Click "Generate Contribution Guidelines" to get AI-powered steps for contributing.

---

🖼️ Architecture Diagram

(If you want to use the Mermaid diagram above, you can render it with Mermaid Live Editor.)

---

🤖 Tech Stack

• Python, Streamlit, MongoDB, SentenceTransformers
• Google Vertex AI (Gemini LLM)
• GitLab API

---

🙌 Team

• LearnSol

---

📄 License

MIT License