Build your own · a starting point
Want a scoring tool like this for yourself?
This engine started from a single written request to Claude, then grew in stages — a live fares feed, a watcher, a real domain — with every irreversible step held for explicit approval along the way. Below is a version of that starting request you can copy, adapt to what you're actually comparing, and paste into a new Claude conversation.
I want to build a personal tool that scores options I'm comparing — flights, routes, whatever the decision actually is — against what matters to me, not just price. Here's the pattern I want it to follow: 1. Let me rank what matters to me in order — cost, comfort, flexibility, whatever my real priorities are — and weight every recommendation by that ranking, shown as an explicit breakdown per factor, never collapsed into one unexplained score. 2. Start with a reasonable estimate model if a live data source isn't connected yet, but always label clearly on-screen which mode I'm in — modeled estimate versus real live data — and never present one as the other. 3. Keep anything sensitive, like payment or account details, entirely local to my browser — never sent anywhere or stored remotely. 4. Publish it as a small, real website I can use from my phone, with the scoring logic running instantly, without waiting on a server round-trip. 5. Let me set up a watch on something I care about, with a scheduled check that re-runs itself and tells me about a meaningful change without my having to ask. 6. Keep the backend radically simple — a feature that needs real infrastructure has to earn it, not get added by default. Please help me scope this into a first small version we can finish today, tell me exactly what you'd need from me at each step (any account access, any purchase like a domain name, anything you can't do without my say-so), and don't do anything irreversible without checking with me first.
How this actually gets maintained
All of this runs inside Cowork, working from a single project folder Claude keeps on your behalf — the code, the data, and a file called CLAUDE.md: Claude's own running notes on how the system works, decisions already made, and what's still open. Every new Cowork session reads that folder first, so it picks up exactly where the last one left off, and writes its notes back before it ends. You never have to re-explain the project from scratch, even months later.
What it takes on your end
- A Claude account with access to Cowork.
- A few short check-ins over the following days or weeks — this kind of tool improves in passes, not one sitting.
- A willingness to say yes or no when Claude asks — it will hold at anything with a real-world consequence rather than guess.
Curious how this specific tool actually works, step by step? Read the Guide. Curious what order it was built in and what's still ahead? Read the Roadmap.