Requests, approvals, and running balances.
PTO usually lives in a spreadsheet one person maintains and everyone else distrusts, or an HR suite that bills per employee per month for what is really a ledger and an approval button. The accrual math drifts, two people book the same week, and "how many days do I have left" becomes a Slack message to HR. This blueprint builds the real thing: balances that add up, requests that route to a manager, a calendar of who is out, all in code you own, not a seat you rent.
A PTO system backed by SQLite, seeded with eight employees on varied accrual rates so every screen has data on load. Employees pick dates and see their projected balance before submitting, with a warning when team requests overlap. Managers get an approval queue: one-click approve or deny, plus a comment. Underneath, a balance engine accrues monthly, applies a carryover cap, and keeps per-employee history. A team calendar shows who is out when, and a year-end report lines up taken against accrued.
You watch it happen in the task workspace: the agent scaffolds the project, builds the schema, seeds pending, approved, and denied requests, and confirms the views render before handing you a live preview. The code lands on a branch in your repository with a pull request ready.
The first version is a base you steer with plain follow-up messages:
Each follow-up wakes the same repository and the agent continues on its own code.
Do I need to know how to code? No. You describe the changes; the agent writes the schema, logic, and views, then ships the code to your repository.
Where does the data live? In your own SQLite database, in your repository. Your team's records stay under your control.
Can it match our accrual rules? Yes. The monthly accrual and carryover cap live in the code, so ask for your exact policy and the agent adjusts the engine.