Budgets and spending reports, built by Claude Code on Elements

Thriftledger is a full-stack app built with Claude Code on Elements.

Bank accounts, CSV import that skips duplicates, rules that categorize, monthly budgets and spending reports.

Run this app on your machine. Use this demo as inspiration, or as the starting point for your own app. Your agent creates a new app with this one as the scaffold, and you prompt it to add whatever features you want.

How it's built

Thriftledger needed a ledger that stays current in every open tab, a CSV importer that knows what it has already seen, rules that sort spending, and budgets that move as transactions arrive. Each of those is a part of Elements, so the agent spent its 24 minutes on the money itself.

• A live ledger. Accounts, categories, rules, budgets, imports and transactions are LiveTables, each opened for the signed-in household. A trigger carries every transaction write to the open pages, so the dashboard's budgets move the moment an import lands in another tab.

• CSV import as a function call. The import page previews the file in the browser, then sends it to an @rpc that parses the rows on the server and inserts them in one transaction. Each row carries a fingerprint, so a re-import skips what is already there, and the column mapping is saved for that bank's next export.

• Rules that categorize. Every imported row takes the longest matching rule, and one rpc runs the rules over everything still uncategorized in a single SQL update. Edits go through the transactions LiveTable, which checks that a split's parts add up to the whole.

• Reports from the data. The reports page draws its spending charts as SVG from the ledger's own rows.

• Data from SQL files. Two migrations define the ledger and seed one household with three accounts, categories, rules, budgets and six months of activity ending today.

Why Elements

Elements is an integrated full-stack app framework and a project server. The project server checks the agent's work as it builds: it type-checks the server and browser code, analyzes the program, applies migrations and runs the tests, then reports the results to the agent. A security mistake, such as calling a server-only function from the browser, is a compile error. The agent reads the report and fixes its own mistakes before it tells you a feature is done. The same checks run on every feature you ask for next.