Georgia Tech Comprehensive MCP Server

An advanced MCP (Model Context Protocol) server that provides LLMs with comprehensive access to Georgia Tech's academic and research ecosystem, integrating multiple GT systems for intelligent workflows. Now with OSCAR 500 error fixes for reliable course searching and ChatGPT HTTP integration.

Features

ChatGPT Integration ✨ NEW: HTTP API Server

• FastAPI HTTP Server: Complete HTTP API for ChatGPT Custom Tools integration

• CORS Enabled: Cross-origin requests from ChatGPT domains supported

• JSON Responses: All endpoints return proper JSON for ChatGPT consumption

• Auto Documentation: Interactive OpenAPI/Swagger docs at /docs endpoint

• Health Monitoring: Real-time service health checks and monitoring

Core Course Scheduling (OSCAR System) ✨ 500 Error Fixes Applied

• Available Semesters: Get list of available semesters for course searches

• Subject Lookup: Get available departments/subjects for a given semester

• Course Search: Search for courses by subject, course number, or title with improved reliability

• Course Details: Get detailed information including seat availability, waitlist info, and restrictions

• Improved Workflow: Fixed 500 server errors by implementing proper GT navigation patterns

Research & Knowledge Systems (SMARTech Repository)

• Research Paper Search: Search 500+ research papers, theses, and publications

• Faculty Research Matching: Find faculty by research interests and collaboration history

• Research Trend Analysis: Analyze publication trends over time

• Cross-Referencing: Link research areas to related courses

Campus Infrastructure (Places & GIS)

• Location Services: Search campus buildings and facilities

• Accessibility Information: Detailed accessibility features and routing

• Service Discovery: Find buildings with specific services (AV equipment, catering, etc.)

• Spatial Analysis: Route planning and proximity searches

Cross-System Integration

• Research-Course Correlation: "What courses support my robotics research?"

• Faculty-Course Matching: "Who teaches courses related to my research area?"

• Resource Optimization: "Find labs with networking equipment near CS building"

• Academic Planning: "Plan degree path with research opportunities"

Setup

Automated Setup (Recommended)

1. Run the setup script:

   ./setup.sh

   This will automatically create the conda environment and install all dependencies.

2. Start the server:

   # For ChatGPT integration (HTTP API server): ./start_server_chatgpt.sh # For EXPANDED functionality (all GT systems): ./start_server_expanded.sh # For original course scheduling only: ./start_server.sh

Manual Setup

1. Create conda environment:

   conda create -n gtmcp python=3.11 -y conda activate gtmcp

2. Install dependencies:

   pip install -r requirements.txt pip install -e .

3. Test the functionality:

   # Test original course scheduling: python test_server.py # Test expanded multi-system functionality: python test_expanded_server.py

4. Run unit tests:

   python -m pytest tests/ -v

5. Run the server:

   # ChatGPT HTTP API server: python -m gtmcp.server_fastapi --host 0.0.0.0 --port 8080 # Expanded server (recommended): python -m gtmcp.server_expanded # Original course scheduling only: python -m gtmcp.server

ChatGPT Integration Setup

Quick Start

1. Start the HTTP server:

   ./start_server_chatgpt.sh

   Server will run on http://0.0.0.0:8080 by default.

2. Configure ChatGPT:

   • Open ChatGPT settings

   • Go to Beta Features

   • Enable "Custom GPTs & Tools"

   • Create new custom tool:

     • Name: Georgia Tech MCP Server

     • Description: Access GT course schedules and research

     • URL: http://localhost:8080

Available HTTP Endpoints

Example ChatGPT Queries

• "What CS courses are available for Spring 2025?"

• "Find research papers about machine learning"

• "Get details for course CRN 12345 in Spring 2025"

• "What subjects are available for Fall 2024?"

MCP Tools (17 Comprehensive Tools)

Course & Academic Tools

get_available_semesters

Get list of available semesters.

• Input: None

• Output: List of semesters with codes and names

get_subjects

Get available subjects/departments for a semester.

• Input: term_code (e.g., "202502" for Spring 2025)

• Output: List of subject codes and names

search_courses

Search for courses in a given semester and subject.

• Input:

  • term_code: Semester code

  • subject: Subject code (e.g., "CS", "MATH")

  • course_num (optional): Course number filter

  • title (optional): Title search filter

• Output: List of matching courses with CRNs

get_course_details

Get detailed information for a specific course.

• Input:

  • term_code: Semester code

  • crn: Course Reference Number

• Output: Detailed course info including seats, waitlist, restrictions

Research & Knowledge Tools

search_research_papers

Search Georgia Tech research repository for papers.

• Input:

  • keywords: Array of search keywords

  • subject_areas: Subject areas to filter by

  • date_from/date_until: Date range filters

  • max_results: Maximum results to return

• Output: List of research papers with abstracts and metadata

find_faculty_research

Find faculty research profiles by research area.

• Input: research_area (e.g., "robotics", "AI")

• Output: Faculty profiles with research interests and publications

analyze_research_trends

Analyze research trends over time for keywords.

• Input:

  • keywords: Keywords to analyze

  • years: Number of years to analyze

• Output: Trend analysis with yearly publication counts

get_repository_info

Get information about the GT research repository.

• Input: None

• Output: Repository metadata and statistics

Campus & Location Tools

search_campus_locations

Search for campus buildings and locations.

• Input:

  • query: Search query for building/location name

  • services: Required services (e.g., "AV equipment")

  • accessible: Filter for wheelchair accessible locations

• Output: List of matching campus locations

get_location_details

Get detailed information about a specific location.

• Input: building_id: Building identifier

• Output: Complete location details including services and accessibility

find_nearby_locations

Find locations near a specific building.

• Input:

  • center_building_id: Building to search around

  • radius_meters: Search radius

  • services: Services to filter by

• Output: List of nearby locations within radius

get_accessibility_info

Get detailed accessibility information for a building.

• Input: building_id: Building identifier

• Output: Comprehensive accessibility features and services

Cross-System Integration Tools

suggest_research_collaborators

Suggest potential collaborators based on research interests.

• Input:

  • research_area: Research area for collaboration

  • keywords: Specific research keywords

• Output: Suggested faculty and researchers with overlap analysis

find_courses_for_research

Find courses related to a specific research area.

• Input:

  • research_topic: Research topic or area

  • term_code: Semester to search in (optional)

• Output: Related courses with research connections

check_system_health

Check health status of all integrated GT systems.

• Input: None

• Output: Status report for OSCAR, SMARTech, Places, and other systems

Configuration

The server can be configured via config.json:

Command Line Options

Usage Examples

Basic Course Scheduling

Research & Faculty Discovery

Campus Navigation & Services

Cross-System Intelligence

Rate Limiting

The scraper includes configurable delays and retry logic to be respectful to the GT OSCAR system:

• Default 1-second delay between requests

• 30-second timeout per request

• Up to 3 retries with exponential backoff

Testing

The project includes comprehensive testing for all integrated systems:

Comprehensive Test Suite (124+ Tests)

• 38 HTTP Server Tests: FastAPI endpoint testing and ChatGPT integration validation

• 17 External Server Tests: Real HTTP server integration testing with subprocess management

• 69+ MCP Unit Tests: Original MCP functionality validation and client testing

HTTP Server & ChatGPT Integration Tests

Unit Tests

• Model Tests: Data validation and serialization

• Configuration Tests: Config loading and validation

• Client Tests: All GT system clients with mocked responses

• Server Tests: MCP server tools and error handling

• Integration Tests: Cross-system workflow validation

Run all tests:

Run specific test modules:

Integration Tests

Test Categories

FastAPI Server Tests (21 tests)

• ✅ Basic Functionality: Root, health, and tools endpoints

• ✅ OSCAR Integration: Semesters, subjects, courses, and details with mocked data

• ✅ Research Integration: Paper search and research endpoints

• ✅ ChatGPT Compatibility: CORS, JSON responses, AI plugin manifest, legal endpoint

• ✅ Performance: Concurrent requests, large responses, error recovery

External Server Tests (17 tests)

• ✅ Real HTTP Server: Subprocess startup and external HTTP testing

• ✅ ChatGPT Integration: Cross-origin requests and response validation

• ✅ Server Reliability: Load testing, memory stability, error recovery

• ✅ API Documentation: OpenAPI spec and interactive docs validation

System Health Checks

Error Handling

The application includes comprehensive error handling:

• Network Errors: Retry logic with exponential backoff

• Parse Errors: Graceful handling of malformed HTML

• Validation Errors: Input validation with clear error messages

• Server Errors: Structured error responses for MCP tools

All errors are logged with appropriate severity levels and include helpful context for debugging.

Note on Data Availability

Georgia Tech may not maintain course details beyond the next semester, even though older semesters appear in the dropdown. Always check that courses are actually available for the requested semester.