# Museance

- **Event:** [AIE Code Summit 2026 Hackathon](https://cerebralvalley.ai/e/aie-code-2026-hackathon)
- **When:** Sat, Nov 7 at 9:00 AM – 10:00 PM (PST)
- **Where:** San Francisco, CA
- **Team:** [Alejandro Delmar](https://cerebralvalley.ai/u/adelmar), [Lokesh lks](https://cerebralvalley.ai/u/Lkslokesh), [Vi Huynh](https://cerebralvalley.ai/u/vihuynh72)
- **GitHub:** https://github.com/vihuynh72/museance
- **Demo video:** https://www.loom.com/share/9e70497c079749f48ad88b72fc065057
- **Gallery:** https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/39

The problem
- Supermarket cold rooms run on old controllers that accept unauthenticated commands. In the Oct/Nov 2026 EU attacks, a single remote write could switch off cooling and silence the alarm.
- Rebuilding that legacy software is risky: an AI can rewrite it fast, but nobody can prove the rewrite keeps the food cold.

What MUSÉANCE does
- AI proposes, the court decides. An AI rebuilds and patches the controller, and swears how every behavior works, with credits on the line.
- A court with no AI tests both builds side by side over Modbus TCP, the controller's own industrial protocol, and runs every check twice.
- Every verdict goes into a hash-chained ledger. Only behaviors that survive the court get locked into the record.

Proof it works (measured today)
- Cooling on and off, temperature hysteresis, the high-temperature alarm and the defrost schedule: 6/6 behaviors identical on both builds, all locked, zero flaky.
- 3/3 unauthorized configuration changes: the legacy controller accepted every one, and the patched build rejected all three, kept troom at 4 °C and its alarm working.
- The patch is three lines. The ledger holds 24 events with the chain intact. The same court also scored a revived 2016 todo app fro41% to 100% faithful.
                                                                                                                                    ilt with
- Fetch.ai agents argue the case live in ASI:One. SubQ scores each claim's confidence. MongoDB's AI finds the evidence behind every claim. Codex and Meta Muse rebuild.

Our mission
- Make AI fixes to critical infrastructure accountable: every claim is staked, every verdict comes from a test, and nothing ships
  unproven. Modeled controller, not vendor code.

## More from AIE Code Summit 2026 Hackathon

- [AgentGlass](https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/36)
- [rsi_today](https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/37)
- [Sisyphus](https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/38)
- [MERGE-GATE](https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/40)
- [KT-coding-agent](https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/41)
- [Doppel](https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/42)

---

Markdown version of https://cerebralvalley.ai/e/aie-code-2026-hackathon/hackathon/gallery/39. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt