Content
# LMS MCP Assistant
AI-powered LMS assistant that monitors assignments, generates solutions, creates formatted submission documents, and submits with user approval.
Built for **SSN College of Engineering LMS** (Moodle-based).
## Architecture
```
LMS (Moodle) ←→ Playwright ←→ MCP Server ←→ AI Agent
↕
FastAPI Web Dashboard
↕
SQLite Database
```
## Quick Start
```bash
# Setup
python -m venv .venv
source .venv/bin/activate
pip install -e .
# Install Playwright browser
playwright install chromium
# Configure credentials
cp .env .env.local # Edit with your LMS credentials & AI API keys
```
## Commands
| Command | What it does |
|---|---|
| `python -m app.main mcp` | Run MCP server (stdio) — connect AI agent |
| `python -m app.main web` | Run web dashboard at http://127.0.0.1:8080 |
| `python -m app.main all` | Run MCP server + web UI + scheduler |
| `python -m app.main init-db` | Initialize/reset the database |
## MCP Tools (12 tools)
| Tool | Description |
|---|---|
| `login()` | Log in to LMS |
| `logout()` | Log out |
| `list_assignments()` | List all visible assignments |
| `get_assignment(id)` | Get full assignment details |
| `download_attachment(id, url, filename?)` | Download attachment file |
| `classify_assignment(id)` | AI classify (doc/programming/handwritten/etc) |
| `solve_assignment(id)` | AI generate solution |
| `generate_document(id, fmt='docx')` | Create formatted DOCX/PDF |
| `upload_submission(id, file_path)` | Upload file to LMS |
| `submit_assignment(id)` | Submit assignment |
| `get_submission_status(id)` | Check submission status |
## Project Structure
```
lms_mcp/
├── app/
│ ├── mcp/ # MCP server + tool registrations
│ │ ├── server.py # FastMCP setup, tool decorators
│ │ └── tools.py # Tool implementations
│ ├── lms/ # Playwright LMS automation
│ │ ├── auth.py # Login/logout
│ │ ├── browser.py # Browser lifecycle management
│ │ ├── assignments.py # List, get, download
│ │ ├── submission.py # Upload, submit, status
│ │ └── selectors.py # Moodle CSS selectors
│ ├── ai/ # LLM integration
│ │ ├── client.py # Anthropic/OpenAI abstraction
│ │ ├── classifier.py # Assignment type classification
│ │ └── solver.py # Solution generation
│ ├── documents/ # Document generation
│ │ ├── generator.py # DOCX and PDF generation
│ │ └── templates.py # Document templates
│ ├── database/ # SQLite persistence
│ │ ├── models.py # Schema
│ │ └── repository.py # CRUD operations
│ ├── notifications/ # User notifications
│ │ └── notifier.py
│ ├── scheduler/ # Periodic monitoring
│ │ └── monitor.py # APScheduler
│ ├── ui/ # Web dashboard
│ │ ├── app.py # FastAPI app
│ │ ├── routes.py # API endpoints
│ │ └── templates/ # Jinja2 templates
│ └── main.py # Entry point
├── config/
│ └── settings.py # Configuration
├── data/ # SQLite database
├── generated/ # Output documents
└── site_files/ # Reference HTML snapshots
```
## Workflow
1. **Monitor** → Scheduler polls LMS for new assignments
2. **Notify** → New assignments are detected and saved
3. **Classify** → AI determines assignment type
4. **Solve** → AI generates solution content
5. **Document** → Generate formatted DOCX/PDF with student info
6. **Review** → User previews/edits/regenerates (via tools)
7. **Submit** → Upload + submit on LMS (user-approved only)
## Environment Variables (`.env`)
```
LMS_URL=https://lms.ssn.edu.in
LMS_USERNAME=your_email@ssn.edu.in
LMS_PASSWORD=your_password
ANTHROPIC_API_KEY=sk-ant-...
STUDENT_NAME=Your Name
STUDENT_REG_NO=2410112
```
MCP Store