# Marco Rotili - Handgemacht.ai

- **Event:** [Built with Opus 4.6: a Claude Code hackathon](https://cerebralvalley.ai/e/claude-code-hackathon)
- **When:** Feb 10 at 12:00 PM – Feb 17 at 10:00 AM (EST)
- **Where:** Location TBA
- **Team:** [Marco Rotili](https://cerebralvalley.ai/u/marot)
- **GitHub:** https://github.com/marot/spotter
- **Demo video:** https://drive.google.com/file/d/1o8eP8mI8sJqZiJIpQIVvKgKHlFfQp8BX/view?usp=drive_link
- **Gallery:** https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery/158

Spotter tries to tackle problems that arise when you work with agent orchestrators on multiple projects at the same time:

1. You forget details about the product. Exact requirements, architecture
2. There is too much code to review

Quality decreases over time.

Spotter provides you with tooling to help you mitigate this with two novel approaches.

First, it moves review left. Instead of reviewing the output of the agent, spotter allows you to attach comments to the agent and subagent transcripts. Using a MCP server, Claude Code can then address those comments and turn them into improvements for Claude Code or follow up issues for product fixes.

Second, to help you remember important facts about the product and transcripts, it implements a spaced repetition system that surfaces information over time. Opus  is used to extract the relevant information out of transcripts and commits.

Together, they build a lens into the inner workings of Claude Code, and into your product.

---

Markdown version of https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery/158. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt