I keep a file in Obsidian called “whats-next.” It’s where tasks, ideas, and random thoughts go to live until I do something about them.
Used to be, I’d open it manually, scan through, prioritize in my head, then jump into whatever felt urgent. Standard workflow.
Now it’s different.
Claude Code has a skill that pulls content from that file when I say something like “hey, what’s next” or “what needs attention.” When I say “take a note” or “remind me to do this,” it writes directly to the file.
On startup, my CLAUDE.md instructs Claude to read the file and surface what needs attention.
The part that makes this useful: it’s just a local text file. I can add to it from my Mac, from my phone, anytime. No app to open. No sync issues. Just text.
I can arrange it by date, by urgency, by project—whatever makes sense that day. I don’t have to fight with a rigid task manager’s interface. I just ask for context and get it.
It’s become a substitute for a traditional to-do list, except more flexible. Need everything sorted by deadline? Done. Want to see only work items? Easy. Looking for that random thought from last Tuesday? It’s there.
The barrier between capturing a thought and acting on it dropped to almost zero. And the barrier between “what should I work on” and “here’s what matters right now” disappeared.
Turns out, the best to-do list is one that talks back.