Build an expense tracker, then repair its filters with focused prompts
At a glance
A developer uses Kiro to generate a browser-based expense tracker and iterates on problems found during manual testing. This workflow is listed under AI-Assisted Development, Expense Tracking, Manual Testing.
- Agents / tools
- Kiro, Web browser
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) With Kiro available in a writable local project and a browser for checking the result, define the requirements: expense amount, category, date, notes, monthly budget, and local browser storage. Have Kiro capture the requirements in Markdown. 2) In Vibe mode, request the tracker as HTML, CSS, JavaScript, and a web manifest. Run the generated page in the browser. 3) Request a desktop layout with a fixed expense-entry sidebar and dashboard. Exercise date filters, category buttons, charts, and dark mode; describe each failure in a separate repair prompt. 4) Recheck the repaired behavior, including totals after filtering and rejection of future expense dates. Finish by generating the architecture, user guide, and prompt-history documents alongside the code.
How it went
The author published a demo and repository and reports fixing overflowing buttons, poor dark-mode contrast, broken filters, and charts through targeted prompts.
Worth knowing
The described app stores data in browser LocalStorage and has no backend.
