Context is king
The highest-leverage thing you can do with AI coding tools isn't writing better prompts. It's giving them better context.
About a year and a half ago, Anthropic announced Claude Code. I was probably one of the first people in Serbia to get access. I'd been following their mailing list and Twitter closely, so when it dropped, I was ready.
At the time, my note-taking setup was a mess. I had notes scattered across Notion and Obsidian. The Obsidian vault was the bigger problem. I had written things down, but I never followed any consistent structure or best practices for organizing them. When it came time to actually review those notes, I'd look at the chaos and just not bother. The lack of structure made everything feel like more effort than it was worth.
Claude Code and markdown
I tried using Claude Code specifically for this: restructuring my Obsidian notes. Even at that early stage, it was surprisingly good at working with markdown. It could take a disorganized set of notes, figure out what mattered, and restructure everything into something I'd actually go back to.
I kept going. I started feeding it the notes I used most frequently, having it clean them up and add proper structure. The vault went from something I avoided to something I actually used.
The real insight
The models have gotten significantly better since then. But the most useful thing I took away from that experience wasn't about what the model could do. It was about what I was giving it.
The more context you provide about what you're working on, what you're about, and what you need help with, the better the output gets. Not slightly better. Fundamentally better.
This applies regardless of which tool you're using. Claude Code, OpenCode, Codex, Gemini CLI, whatever comes next. The context you provide, your project docs, your architecture notes, your AGENTS.md, is the single highest-leverage input you control.
Better prompts help. Better context changes the quality of the work entirely.