AI skill files can be borrowed, but personalization still requires your own corrections
Understand this story
In his article, Andrea Azzini suggests that other people's AI skill files are like an Italian grandmother's ragù recipe: you can read and borrow from them, but the parts that truly work well come from the user's own repeated corrections. Using the construction of Triagon, a front-desk request triage agent, as an example, he says that through small corrections made one at a time, the responses became more accurate and better matched his own tone, and each correction was retained and accumulated. Azzini believes that downloading someone else's skill can save a week of time, but skipping the "cooking it yourself" step means you never develop the muscle for personalized configuration.
AI overview based on source reports · Updated 30 min ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AINonna's RAGù: AI skill files only get you started
Andrea Azzini argues that downloaded AI skill files are like a family ragù recipe: readable and tweakable, but the value comes from your own repeated corrections over time. He describes building Triagon, an agent that triages front desk requests, where each small fix to context-gathering, when to reply, and voice compounded week over week. He concludes that personalising your setup is a muscle trained only by iterating on your own problems, and that returns arrive later and larger than expected.
Activity for this story
Not enough continuous observations to show a trend yet.