Git Conflict Challenge

During development of Sharpner, we frequently ran into Git merge conflicts, especially when multiple contributors were working on the same core files (editor logic, Socket.io handlers, and API routes).

The Issue

• Parallel feature development caused conflicts in shared files
• Rebase and merge operations occasionally overwrote important changes
• Conflicts became harder to resolve as the codebase grew

How We Solved It

• Broke large files into smaller, modular components
• Followed a feature-branch workflow instead of committing directly to

  main

• Used clear commit messages and rebased frequently
• Manually reviewed conflicts line-by-line to avoid silent bugs

What We Learned

• Good Git practices scale just as much as good code
• Modular architecture drastically reduces merge conflicts
• Frequent syncing prevents painful late-stage conflicts

This experience helped us improve both our collaboration process and codebase structure.

AI System Instruction Challenge

While integrating the AI assistant in Sharpner, one major hurdle was designing effective system instructions that kept responses helpful, consistent, and safe.

The Issue

• Early prompts produced generic or overly verbose answers
• The AI sometimes ignored context from the editor
• Inconsistent tone and formatting across responses

How We Solved It

• Created a strict system instruction layer that defined:
  • Role (pair programmer + tutor)
  • Response format (Markdown + syntax highlighting)
  • Behavior rules (explain why, not just what)
• Injected editor context (current code, language, error output) into each request
• Iteratively refined prompts by testing real user queries

What We Learned

• Prompt design is as important as model choice
• Clear constraints lead to more reliable AI behavior
• Context-aware instructions drastically improve learning outcomes

This refinement turned the AI from a basic chatbot into a true coding assistant inside Sharpner.