# LexHarmoni

- **Event:** [Built with Opus 4.7: a Claude Code hackathon](https://cerebralvalley.ai/e/built-with-4-7-hackathon)
- **When:** Apr 21 at 12:00 PM – Apr 27 at 2:00 AM (EDT)
- **Where:** Online
- **Team:** [Ziffany Firdinal](https://cerebralvalley.ai/u/Ziffan)
- **GitHub:** https://github.com/ziffan/lexharmoni
- **Demo video:** https://youtu.be/v1EbVazszEs
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/18

LexHarmoni is a stress-test harness for legal drafting — the same discipline software engineers apply when they test code in staging before shipping to production, applied to Indonesian financial services regulations (POJK and SEOJK, the rule-making instruments of OJK, Indonesia's financial regulator). It detects friction across overlapping regulatory documents — normative conflicts, hierarchical orphans, and terminology drift — that emerge not from negligence, but from sheer accumulation across years of layered rule-making.

Built by a former regulator, the tool uses Claude Opus 4.7 with full-context reasoning over a corpus of seven regulations, active and historical. A 1-hour prompt cache keeps inference cost at roughly $1.70 per warm run, making pre-enactment stress-testing operationally viable — a regulatory body can run dozens of draft revisions per month for the cost of a single staff hour. A dual-stream UI surfaces both findings and the model's reasoning trail, so every claim is auditable against article-level citations.

Validated against a manual baseline, three consecutive runs surfaced the same ground-truth frictions — including a 19-month collection-hours mismatch between two active rules — with zero hallucinated citations. The repository ships under Apache 2.0 with full documentation, including a replication guide for non-engineers.

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/18. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt