I recently hit a wall, with cursor struggling to resolve a bug in a particular library. It kept offering basic fixes that just weren't right and we kept looping through the same fixes burning throught tokens
Thats when i remember what I did back in the days when we all coded first and googled second
The Problem:
Modern AI models are powerful, but their training data has a knowledge cutoff. This means they do not have real-time information on the absolute latest bugs, edge cases, or workarounds being discussed and solved right now in the open-source community.
The Solution: tell your AI to search GitHub issues.
Instead of asking "How do I fix X?", you prompt it to leverage GitHub issues. This shifts the AI from trying to recall (potentially outdated) information to actively searching a live, community-maintained database of real issues from real developers
The Prompt I Used:
this bug in [Library/Framework Name, e.g., 'React Query', 'TensorFlow', 'MyProjectName'] can you check recent GitHub issues for related solutions or workarounds? Specifically, look for issues matching [specific error message, function name, or keyword].
Augmenting the intelligence of the AI tool with the intelligence of the community taps you into the collective intelligence of thousands of developers who have likely encountered (and solved) similar issues.
Have you tried anything similar? Share your AI prompting hacks in the comments!
Top comments (0)