Hello
I have been using cursor in my organisation for past 6 months now. In earlier days, I was able to use it very well using the ask and plan modes. As the features became more complex and the vast codebase became a blocker. Few of the problems I started facing -
- every chat started with no context of what we have build till now
- Agent used to omit details, hallucinate and write buggy code
- token usage went off the roof
in order to mitigate this, i came across few people who have built full SDLC using skills. here are some of them i found interesting -
- plugins/pstack at main · cursor/plugins · GitHub
- GitHub - mattpocock/skills: Skills for Real Engineers. Straight from my .agents directory. · GitHub
- GitHub - addyosmani/agent-skills: Production-grade engineering skills for AI coding agents. · GitHub
These are obviously very well designed but again cant fit my use case exactly. I want to learn how can i built my own SDLC using cursor by able to trust AI more with gaurdrails. Is there any resource on the deep learning platform which can guide me for this