# OmniBridge

- **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
- **Placement:** Finalist
- **Team:** [Adinda Panca](https://cerebralvalley.ai/u/adindamochamad)
- **GitHub:** https://github.com/adindamochamad/omnibridge
- **Demo video:** https://youtu.be/RX-uGdIQLLU
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/51

OmniBridge is an AI agent that identifies unknown serial device protocols autonomously — a workflow that didn't exist a year ago because it needs Opus 4.7's specific capabilities to work.

Plug in a 1980s weighing scale, a Modbus PLC, or an Arduino sensor. OmniBridge runs Claude as a multi-step agent with six purpose-built tools — for binary streams, one surgical call to analyze_binary_structure validates CRC16 across every frame and returns frame structure, enough evidence to commit with high confidence. Every reasoning step streams live. Once identified, Claude-generated regex (ASCII) or byte-offset extractors (binary) run locally at thousands of packets per second, zero ongoing API cost.

Traditional protocol integration: engineer reads vendor docs, writes custom parser, ships glue code — eighteen hours median per device. OmniBridge: five minutes. For $1.2 trillion of legacy industrial hardware stranded outside the cloud, that speedup opens entire new markets.

This couldn't work with a smaller model. Adaptive thinking lets Claude spend more compute on binary vs ASCII. Multi-turn tool use with preserved reasoning lets the agent pick a surgical tool and commit. Prompt caching delivers ~65% cache hit on repeat investigations,
keeping cost near $0.20 per identification — roughly one-thousandth of the engineer time replaced.

Open source, MIT licensed, native desktop app on macOS.

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/51. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt