Mastering Claude

# Installation steps:

1. Go to claude.ai/download and download the desktop app
2. You need a Pro account ($20/month, or $17/month billed annually)
3. Open the app and click the "Cowork" tab at the top
4. Select a folder from your computer - this is how Claude reads your files

# Pro tip: Create markdown (.md) files about yourself, your projects,
# your writing style, your brand rules - anything you want Claude to know.
# Point Cowork to the folder containing these files.

# Your first Cowork prompt (use this pattern for everything):

I want to [YOUR TASK] so that [WHAT SUCCESS LOOKS LIKE].

First, read the uploaded files completely before responding.

DO NOT start executing yet. Instead, ask me clarifying questions
so we can refine the approach together step by step.

Only begin work once we've aligned.

# How to set up Opus + Extended Thinking:

1. Open any Claude chat (on claude.ai or Cowork)
2. Click the model selector dropdown at the bottom of the chat
3. Select "Opus 4.6"
4. Turn on "Extended Thinking"

# For API usage:
import anthropic

client = anthropic.Anthropic()

response = client.messages.create(
    model="claude-opus-4-6",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000  # How much "thinking" to allow
    },
    messages=[{
        "role": "user",
        "content": "Analyze the trade-offs between microservices and monoliths for a team of 5."
    }]
)

# The response includes both thinking and final answer
for block in response.content:
    if block.type == "thinking":
        print("Claude's reasoning:", block.thinking)
    elif block.type == "text":
        print("Final answer:", block.text)

# Setting up Connectors:

1. Go to Settings > Connectors
2. Browse the directory of available integrations
3. Click "Add" on the ones you want
4. Authorize access when prompted

# Once connected, you can ask Claude things like:
# - "Search my Slack for the latest discussion about the API redesign"
# - "Find the Q4 revenue report in my Google Drive"
# - "What did the team decide about the deployment strategy in Notion?"

# Connectors are free on all Claude plans.

# Installing Claude in Excel:

# For individuals:
1. Navigate to the Claude in Excel listing on Microsoft Marketplace
2. Click "Get it now" to install the add-in
3. Open Excel, activate the add-in, and sign in with your Claude account

# For admins (organization-wide deployment):
1. Visit the Microsoft 365 Admin Center
2. Navigate to Settings > Integrated apps > Add-ins
3. Search for "Claude by Anthropic for Excel" in Microsoft AppSource
4. Deploy the add-in to your organization or specific users

# After installation:
# - Mac: Tools > Add-ins to activate Claude
# - Windows: Home > Add-ins to activate Claude
# - Sign in with your Claude credentials and start working

# Requires a paid Claude plan (Pro, Max, Team, or Enterprise).
# Official docs: https://support.claude.com/en/articles/12650343-use-claude-in-excel

# Prompt this in Claude Cowork:

Create a professional Excel spreadsheet (.xlsx) for: [PURPOSE]
Context: [WHO IS IT FOR / HOW WILL IT BE USED]
It should cover: [LIST WHAT YOU WANT TO TRACK OR CALCULATE]

Rules:
- Use Excel formulas (SUM, SUMIF, IF, etc.) - never hardcoded calculations
- Put editable assumptions in their own labeled cells
- Freeze top row, auto-fit columns
- [ANY EXTRAS: charts, dropdowns, conditional formatting, specific currency, etc.]

# Claude generates a complete, formula-driven spreadsheet
# with proper structure, formatting, and calculations.
# Nothing is hardcoded - change an assumption cell and everything recalculates.

# Try these in any Claude chat - no setup needed:

# 1. Interactive tool
"Create an interactive HTML calculator that converts monthly expenses
into annual projections. Include fields for rent, groceries,
transportation, subscriptions, and a total that updates in real time."

# 2. Visual comparison
"Create a visual comparison chart of PostgreSQL vs MySQL vs SQLite
with a clean, modern design."

# 3. Project tracker
"Build me a simple project tracker with columns for Task, Owner,
Status, and Due Date. Make it interactive."

# 4. Architecture diagram
"Make an SVG diagram showing a microservices architecture with
API gateway, auth service, user service, and notification service."

# Claude renders the artifact live in the chat.
# You can interact with it, ask for changes, and download the result.

# Setting up Projects (takes 5 minutes):

1. Go to claude.ai and log in (Pro or Team plan required)
2. Click "Projects" in the left sidebar
3. Click "Create Project" - give it a name (e.g., "API Redesign")
4. Click "Add content" - upload your key files:
   - Brand docs, style guides
   - Architecture documents
   - API specifications
   - Writing samples, reference material
5. Start a new chat inside the project

# Every conversation in this project now has access to all uploaded files.
# Claude references them automatically when relevant.

# Pro tip: Don't overbloat projects with too many files.
# Be selective about what's truly relevant to the project scope.

# Install Claude Code (requires Node.js 18+):
npm install -g @anthropic-ai/claude-code

# Navigate to your project and start:
cd your-project/
claude

# Claude Code will:
# 1. Analyze your project structure
# 2. Read your CLAUDE.md file (if present) for project-specific instructions
# 3. Start an interactive session where you can give it tasks

# Example first interaction:
> Explain the architecture of this project
> Find and fix the bug in the authentication middleware
> Add unit tests for the UserService class
> Refactor the database layer to use connection pooling

# CLAUDE.md - Example for a Python web API project

## Project Overview
FastAPI-based REST API for inventory management.
Python 3.12, PostgreSQL 16, Redis for caching.

## Development Commands
```bash
pip install -e ".[dev]"        # Install with dev dependencies
pytest tests/ -v               # Run all tests
pytest tests/unit/ -v          # Run unit tests only
alembic upgrade head           # Apply database migrations
uvicorn app.main:app --reload  # Start dev server
```

## Architecture
- app/main.py        -> FastAPI app entry point
- app/routers/       -> API route handlers
- app/models/        -> SQLAlchemy models
- app/services/      -> Business logic layer
- app/repositories/  -> Database access layer
- tests/             -> Pytest test suite

## Coding Standards
- Use Pydantic v2 syntax (model_config, not class Config)
- All endpoints must have type hints and docstrings
- Use dependency injection for database sessions
- Never use datetime.utcnow() - use datetime.now(timezone.utc)

## Important Notes
- Never commit .env files
- Always run tests before committing
- Use conventional commits (feat:, fix:, docs:, etc.)

# .claude/settings.json - Hook configuration example

{
  "hooks": {
    "afterEdit": [
      {
        "command": "npx prettier --write $FILE",
        "description": "Auto-format edited files with Prettier"
      }
    ],
    "afterWrite": [
      {
        "command": "npx eslint --fix $FILE",
        "description": "Auto-lint new files"
      }
    ]
  }
}

# Hooks fire automatically - Claude Code respects their output.
# If a hook fails, Claude Code sees the error and can adjust.

# .claude/settings.json - MCP server configuration

{
  "mcpServers": {
    "postgres": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-postgres"],
      "env": {
        "DATABASE_URL": "postgresql://user:pass@localhost/mydb"
      }
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_TOKEN": "ghp_..."
      }
    }
  }
}

# With MCP servers configured, Claude Code can:
# - Query your database directly to understand schema and data
# - Create GitHub issues and PRs programmatically
# - Access any tool exposed via MCP

# .claude/commands/review-security.md
# Invoked with: /review-security

Review the changes in this diff for security vulnerabilities.

Check for:
- SQL injection risks (unparameterized queries, raw string interpolation)
- XSS vulnerabilities (unsanitized user input rendered as HTML)
- Secrets or credentials hardcoded in the code
- Insecure deserialization
- Missing authentication or authorization checks on new endpoints

For each issue found:
1. Show the vulnerable code snippet
2. Explain the risk
3. Provide a fixed version

# .claude/commands/add-tests.md
# Invoked with: /add-tests

Generate a comprehensive pytest test suite for the file or function
I describe. Follow these project conventions:

- Use pytest fixtures defined in tests/conftest.py
- Mock external services with unittest.mock.patch
- One test file per source file, mirroring the directory structure
- Test names: test_<function>_<scenario> (e.g. test_create_user_duplicate_email)
- Include happy path, edge cases, and failure scenarios
- Assert both the return value and any side effects (DB writes, events fired)