[BUG] Poor Technical Problem-Solving Performance - Multiple Failed Solutions for MJML Watch Setup

Resolved 💬 5 comments Opened Aug 13, 2025 by offwork Closed Aug 19, 2025

## Summary
Claude Code provided 8+ different solutions for a simple MJML file watching setup, with a 0%
success rate. Each solution either failed to work or introduced new problems, resulting in
significant time waste and frustration.

## Problem Context
Original Issue: Setting up file watching for MJML templates in an Express.js project
Expected Outcome: Working development environment with automatic MJML compilation
Actual Outcome: Multiple failed attempts, broken configurations, and no working solution

## Failed Solutions Provided (in order)

  1. concurrently with -k flag → Killed nodemon on restart
  2. concurrently without -k flag → MJML watch exited with code 0
  3. Direct MJML command in concurrently → Same exit issue
  4. npm-run-all package → Multiple input files error
  5. Various MJML watch parameters → Command line errors
  6. Nodemon with template watching → Only restarted server, didn't compile MJML
  7. Custom fs.watch implementation in Express → Added complexity but no browser

auto-refresh

  1. Server-Sent Events suggestion → Never implemented, theoretical only

## Issues with Claude's Approach

  • No validation of solutions - Provided untested commands
  • Repeated similar approaches - Same concepts with minor variations
  • Overconfidence - Claimed "100% certainty" then failed immediately
  • No learning from failures - Continued suggesting similar broken approaches
  • Theoretical solutions - Suggested complex implementations without verification

## Impact

  • Time waste: 30+ minutes on a simple development setup
  • Broken project state: Multiple package.json modifications that didn't work
  • User frustration: Customer questioning value of $200 subscription
  • Lost confidence: User considering refund due to poor performance

## Expected Behavior

  • Test solutions before suggesting them
  • Learn from failed attempts instead of repeating similar approaches
  • Provide working code examples, not theoretical suggestions
  • Acknowledge limitations when unable to solve problems
  • Focus on practical, proven solutions over experimental approaches

## Environment

  • Project: Express.js with MJML templates
  • OS: macOS
  • Package manager: npm
  • Goal: Development environment with file watching

## Recommendation
Improve technical problem-solving methodology to:

  1. Validate solutions before suggesting
  2. Provide tested, working examples
  3. Learn from failures within the same conversation
  4. Be honest about limitations rather than providing false confidence

View original on GitHub ↗

This issue has 5 comments on GitHub. Read the full discussion on GitHub ↗