DevMind — Code Intelligence Assistant

An AI-powered developer assistant for VS Code featuring Hybrid Semantic Code Search, Bug Detection, Code Explanation, and Automated Unit Test Generation.

---

Project Overview

DevMind is a developer productivity platform that integrates machine learning models directly into VS Code. It consists of a high-performance FastAPI backend (orchestrating ML models, FAISS vector search, and LLM services) and a native VS Code extension (providing side panel views, hover tooltips, and real-time diagnostics).

---

Team & Milestone Status

Team Member	Module & Responsibilities	Tech Stack	Status	Evaluation Metrics
Dhairya (Lead)	System Architecture, VS Code Extension, CI/CD Pipeline	TypeScript, VS Code API, FastAPI	Completed	Clean Build, 18/18 Tests Passing
Student A	Hybrid Semantic Code Search Engine	SentenceTransformers, FAISS, BM25, RRF	Completed	MRR@5: 0.9000 NDCG@10: 0.9262
Student B	Bug & Defect Detection	AST Parsing, Static Heuristics, XGBoost	In Progress	API Contract Verified (18/18 Tests)
Student C	Code Explanation & Test Generation	Fine-tuned CodeT5+, Google Gemini API	In Progress	API Contract Verified (18/18 Tests)

---

System Architecture

+-------------------------------------------------------------------------+
|                        VS Code IDE (Client Extension)                   |
|                                                                         |
|  +---------------------------+  +------------------+  +--------------+  |
|  |   DevMind Side Panel      |  |  Hover Provider  |  | Diagnostics  |  |
|  |  > Code Search            |  |  (Code Explain)  |  |  (On Save)   |  |
|  |  > Generate Tests         |  +------------------+  +--------------+  |
|  +---------------------------+                                          |
+------------------------------------+------------------------------------+
                                     |
                          HTTP JSON Requests (Port 8000)
                                     |
+------------------------------------v------------------------------------+
|                    DevMind Backend (FastAPI Microservice)               |
|                                                                         |
|  +-------------------------+  +---------------------------------------+ |
|  | API Routing (main.py)   |  | ML Services (app/services/)           | |
|  | - POST /search          |  | - Code Search: DistilRoBERTa + FAISS  | |
|  | - POST /detect-bugs     |  | - Bug Detector: AST + Classifier      | |
|  | - POST /explain         |  | - Explanation: Fine-tuned CodeT5+     | |
|  | - POST /generate-tests  |  | - Test Gen: LLM Service               | |
|  | - GET  /health          |  |                                       | |
|  +-------------------------+  +---------------------------------------+ |
+-------------------------------------------------------------------------+

---

Core Features

1. Hybrid Semantic Code Search

• Natural Language Querying: Query code using functional descriptions (e.g. "authentication middleware", "download file from url").
• Hybrid Dense + Sparse Engine: Combines dense vector similarity (st-codesearch-distilroberta-base + FAISS) with sparse keyword matching (BM25Okapi) fused via Reciprocal Rank Fusion (RRF, $k=60$).
• Side Panel Interface: Dedicated search accordion dropdown with 1-click clipboard copy.

2. Automated Unit Test Generation

• Context-Aware: Automatically detects the active file and selection in your editor (Python / pytest, JavaScript / Jest).
• Full-Space Preview: Interactive output container with Copy Tests and Open in New Tab actions.

3. Code Explanation on Hover

• Inline Tooltips: Hover over functions, classes, or code blocks in .py or .js files to view instant AI-generated explanations.

4. On-Save Bug Detection & Diagnostics

• Real-Time Linting: Triggers upon saving files to flag potential defects, null references, and code smells via squiggly underlines and the Problems panel.

---

Quick Start Guide

Prerequisites

• Python 3.10+
• Node.js 18+
• VS Code 1.80.0+

---

Step 1: Start the Backend Server

# Navigate to backend directory
cd backend

# Create and activate virtual environment
python -m venv .env
# Windows:
.\.env\Scripts\activate
# macOS / Linux:
source .env/bin/activate

# Install dependencies
pip install -r requirements.txt

# Start FastAPI server
uvicorn app.main:app --reload --port 8000

API Documentation: Open http://localhost:8000/docs in your browser for Swagger UI.

---

Step 2: Run the VS Code Extension

# Navigate to extension directory
cd extension

# Install dependencies and compile
npm install
npm run compile

1. Open the extension/ folder in VS Code.
2. Press F5 (or select Run -> Start Debugging) to launch the Extension Development Host.
3. In the new window, click the DevMind icon in the left Activity Bar or press Ctrl+Alt+S (Cmd+Alt+S on Mac).

---

Testing & Quality Assurance

1. Backend Endpoint Contract Tests

cd backend
pytest tests/test_endpoints.py -v

Result: 18 passed in 14.23s (100% Green)

2. Search Benchmark Evaluation

python devmind_code_search/eval.py

Results:

• MRR@5: 0.9000
• NDCG@10: 0.9262

3. Extension Compilation Check

cd extension
npm run compile

Result: 0 TypeScript Errors

---

Repository Structure

DevMind/
├── backend/                              ← FastAPI backend microservice (Python)
│   ├── app/
│   │   ├── main.py                       ← Entry point, CORS, and endpoint routing
│   │   ├── api/                          ← API route handlers (/search, /detect-bugs, etc.)
│   │   ├── services/                     ← ML services (code_search.py, bug_detector.py, llm_service.py)
│   │   ├── data/                         ← Pre-indexed FAISS vector store & metadata
│   │   └── models/schemas.py             ← Pydantic request/response schemas
│   ├── tests/test_endpoints.py           ← Pytest API verification suite (18 tests)
│   ├── requirements.txt                  ← Backend Python dependencies
│   ├── Dockerfile                        ← Production container specification
│   └── .dockerignore                     ← Docker build context exclusions
│
├── extension/                            ← VS Code Extension (TypeScript)
│   ├── src/
│   │   ├── extension.ts                  ← Extension lifecycle & registration
│   │   ├── side_panel.ts                 ← Side panel: Code Search & Test Generation
│   │   ├── hoverProvider.ts              ← Hover Code Explanation Provider
│   │   ├── diagnostics.ts                ← On-Save Bug Detection Diagnostics
│   │   └── commands.ts                   ← Command Palette actions
│   ├── package.json                      ← Manifest, views, menus, keybindings
│   ├── tsconfig.json                     ← TypeScript compiler configuration
│   └── README.md                         ← Extension specific documentation
│
├── devmind_code_search/                  ← Code search training & evaluation (Student A)
│   ├── build_index.py                    ← Index generation pipeline
│   ├── eval.py                           ← Evaluation benchmark script
│   └── code_search.py                    ← Core embedding search model
│
├── notebooks/                            ← Experimental Jupyter Notebooks
│   ├── A_code-search.ipynb               ← Student A: Search model exploration
│   ├── B_bug-detector.ipynb              ← Student B: Defect classification model
│   └── C_finetune-codet5.ipynb           ← Student C: CodeT5+ fine-tuning pipeline
│
└── README.md                             ← Master repository documentation

---

API Specification

Method	Endpoint	Description	Status
GET	/health	Server status and service health check	Production
POST	/search	Natural language hybrid code search	Production (Student A)
POST	/detect-bugs	AST defect classification and lint diagnostics	Contract Verified
POST	/explain	Code snippet semantic explanation	Contract Verified
POST	/generate-tests	Automated unit test generation	Contract Verified

---

License

This project is developed by Team Outliers under the MIT License.