ai mock

Inspiration

I wanted to go beyond simple “vibe coding” and properly explore Kiro’s spec-driven development workflow during Kiroween. The goal was to take an idea that would normally stay in a notebook and turn it into a working application with clear requirements, design, and implementation steps—using Kiro’s agentic features to keep the process structured and reviewable.

What it does

[Replace this paragraph with a clear description of your project. Example: “A lightweight tool that revives an old local CLI utility by wrapping it in a modern web interface, adding real-time status and simple sharing.”]

How we built it

• Used Kiro Specs to generate and iterate on requirements.md, design.md, and tasks.md before writing most of the code.
• Applied Steering documents so the agent followed project conventions and architectural decisions consistently.
• Leveraged Agent Hooks to automate formatting, linting, and basic tests on file save.
• Worked primarily in the Kiro IDE, with occasional use of the CLI for headless runs.
• Committed the .kiro directory so the spec and steering history is visible in the repository.

Challenges we ran into

• Balancing the level of detail in the initial specs so the agent had enough guidance without over-constraining creative solutions.
• Getting hooks and steering to work reliably across different file types and sub-folders.
• Keeping the demo polished while still showing the development process and Kiro usage clearly.

Accomplishments that we're proud of

• Produced a working prototype within the hackathon window using a structured, spec-first approach.
• Successfully demonstrated meaningful use of Kiro’s core features (specs, steering, hooks) rather than treating it as a pure chat coding tool.
• Kept the codebase readable and the .kiro artifacts clean enough for judges to review.

What we learned

Spec-driven development with Kiro significantly reduces thrash compared to pure prompt-and-iterate coding. Writing clear requirements and tasks first makes the agent more effective and makes the final project easier to understand and extend. The combination of steering + hooks is especially powerful for maintaining consistency during a short, intense build.

What's next for [Project Name]

• Improve the user interface and error handling.
• Add more MCP integrations for external tools.
• Expand the project based on feedback from the Kiroween community.

Built With

• ai
• java
• python