CLAUDE.md settings.json commands/

Slide 1

CLAUDE.md settings.json commands/

CLAUDE.md settings.json commands/ hooks MCP servers Learn CLAUDE CODE MANTHAN PATEL

Slide 2

CLAUDE.md Your AI's instruction manual. You define: • Project rules & conventions • Code style preferences • Architecture decisions Without CLAUDE.md, Claude is just guessing. 1 # TiltIt -- Manthan's Project Instructions 3 ## Code Standards & Conventions 4 - TypeScript strict mode, no `any` types allowed 5 - Use @phosphor-icons/react for all icons 6 - Dark mode default, Light mode must work from day 1 7 - Semantic Tailwind tokens (text-foreground, bg-muted) 9 ## Architecture & 3D Rendering Engine 10 - Three.js + React Three Fiber for 3D scene rendering 11 - Zustand + subscribeWithSelector for global state 12 - WebCodecs API + Mediabunny for MP4 export pipeline 13 14 ## Commit Format (mandatory after every change) 15 - TI-XXX: Clear description of change (YYYY-MM-DD) 16 - Co-Authored-By: Manthan Patel 17 18 ## Never push unless Manthan says so. No exceptions.

Slide 3

settings.json Control what Claude can do. You define: • Allowed & denied tools • Permission settings • MCP server connections Your guardrails, your rules. settings.json 1 { 2 "permissions": { 3 "allow": , 4 "deny": 5 }, 6 "allowedTools": [ 7 "Edit","Grep","Glob","Read","Write","Bash","Agent" 8 ], 9 "env": { 10 "ANTHROPIC_MODEL": "claude-opus-4-6-20250415", 11 "CLAUDE_CODE_MAX_TURNS": "100", 12 "CLAUDE_CODE_USE_BEDROCK": "0", 13 "CLAUDE_USER": "manthan-patel" 14 }, 15 "mcpServers": { "supabase": { "command": "npx" } }, 16 "autoUpdaterStatus": "disabled", 17 "theme": "dark" 18 }

Slide 4

commands/ Your custom slash commands. You define: • Reusable prompts • Team workflows • One-word shortcuts Type once, use forever. review.md 1 --- 2 description: Review Manthan's code for bugs and security 3 allowed-tools: Read, Grep, Glob, Agent, WebSearch 4 --- 6 Review Manthan's latest changes for these issues: 7 - Security vulnerabilities (XSS, SQL injection, CSRF) 8 - Performance bottlenecks and unnecessary re-renders 9 - Best practice violations and anti-patterns 10 - Missing error handling and edge cases 11 - Type safety issues and implicit any usage 12 13 Focus specifically on: $ARGUMENTS 14 15 Output a markdown table with columns: 16 | File | Line | Severity | Issue | Suggested Fix | 17 Sort by severity: critical > high > medium > low

Slide 5

hooks Automate before & after. You define: • Pre/post tool triggers • Auto-format on save • Shell scripts on events Set it once, runs every time. 1 { 2 "hooks": { 3 "PreToolUse": [{ 4 "matcher": "Edit|Write", 5 "command": "npm run lint -- --fix --quiet" 6 }], 7 "PostToolUse": [{ 8 "matcher": "Write", 9 "command": "npx prettier --write $FILE_PATH" 10 }, { 11 "matcher": "Bash", 12 "command": "echo Manthan: done $(date +%H:%M)" 13 }], 14 "Notification": [{ 15 "command": "terminal-notifier -title Manthan" 16 }] 17 } 18 }

Slide 6

MCP servers Extend Claude's toolbox. You define: • Connect external APIs • Database access • Custom integrations Give Claude superpowers. settings.json 1 { 2 "mcpServers": { 3 "supabase": { 4 "command": "npx -y @supabase/mcp-server", 5 "args": 6 }, 7 "brave-search": { 8 "command": "npx -y @anthropic/brave-mcp", 9 "env": { "BRAVE_API_KEY": "BSA_xxx" } 10 }, 11 "postgres": { 12 "command": "npx -y @anthropic/pg-mcp", 13 "args": 14 }, 15 "github": { 16 "command": "npx -y @anthropic/github-mcp" 17 } 18 }

Slide 7

Want this folder setup? Comment "CLAUDE"