CLAUDE.md — AI-Optimized Documentation for Claude Code

This file provides comprehensive guidance to Claude Code (claude.ai/code) when working with this repository. It emphasizes technical continuity, autonomous development workflows, and architectural understanding.

Critical Context for AI Sessions

Autonomous Development Workflow

Available Scripts

• yarn dev (port 3001) - Development server for human developers
• yarn dev-ai (port 3002) - Development server for AI development
• yarn build - Create production build
• yarn start (port 3001) - Production server for human developers
• yarn start-ai (port 3002) - Production server for AI development
• yarn lint - Run linting checks

Script Execution Pattern

CRITICAL: Always use ONLY these specific log filenames to prevent git pollution:

• ai-output.log - For all build, lint, and one-time commands
• ai-output-start.log - For server commands that need to be killed

DO NOT create descriptive log names like build-dashboard.log or lint-fix.log

# For linting and building (reuses same file)
yarn lint > ai-output.log 2>&1
yarn build > ai-output.log 2>&1

# For dev/start servers (use timeout to prevent hanging)
yarn dev-ai > ai-output.log 2>&1 & sleep 10; kill $!
yarn start-ai > ai-output-start.log 2>&1 & sleep 10; kill $!

After running any command:

1. Read the output log file
2. Parse for errors, warnings, or relevant messages
3. Fix any issues found
4. Re-run the command to verify fixes (overwrites the same log file)
5. Document any new patterns or solutions

Note: The log files are in .gitignore to keep the repository clean

Production Readiness Checklist

• yarn lint passes with no errors
• yarn build completes successfully
• yarn start-ai starts cleanly (after successful build)
• All TypeScript errors resolved