# Day 1 — Claude Code Mastery & Context Engineering
## Lab 01: Project Manager MCP Server

This repository is a submission-ready implementation of the Day 1 MCP lab.

It contains:

- A Python MCP server exposing exactly 3 tools:
  - `create_task`
  - `list_tasks`
  - `update_task`
- SQLite persistence
- Local stdio transport
- Remote SSE transport for Railway/Render
- Claude Code integration via `.mcp.json`
- A `.claude/` configuration with permissions, commands, and an auto-test hook
- A concise project `CLAUDE.md`
- Automated database tests
- Deployment configuration
- Demo/evidence walkthrough
- Context-engineering workshop notes
- Optional extension design notes

## Quick Start

### 1. Install dependencies

```bash
uv sync
```

### 2. Run tests

```bash
uv run pytest -q
```

### 3. Configure Claude Code

Generate `.mcp.json` using the current absolute project path:

```bash
uv run python scripts/configure_mcp.py
```

Verify:

```bash
claude mcp list
```

### 4. Start/use with Claude Code

Open Claude Code in this repository and try:

1. `Create a task called "Set up CI/CD pipeline" with high priority`
2. `List all tasks`
3. `Update task 1 to in_progress`
4. `Show me all high priority tasks that are still todo`

## Remote SSE Deployment

The lab specifically asks for SSE deployment. Run locally with:

```bash
uv run python server_sse.py
```

Endpoints:

- `GET /sse`
- `POST /messages/`

For Railway:

```bash
railway login
railway init
railway up
```

Set `DB_PATH` in the Railway environment.

## Important Evidence Step

The source lab requires a recorded demo or screenshot walkthrough and end-to-end testing in Claude Code. This repository prepares the full implementation and an evidence checklist, but you still need to run Claude Code and capture the final screenshots/video in your own authenticated environment.