started · updated
Developer uses AI and Arduino to pass emissions test for 1994 Jeep
Muhammad Zulali, an AI enthusiast, successfully used Claude and Codex AI models alongside an Arduino Nano microcontroller to help a 1994 Jeep Cherokee pass California emissions testing.
The vehicle had repeatedly failed tests due to nitrogen oxide emissions exceeding limits at specific speeds. Because the 32-year-old engine computer lacked modern OBD2 ports and did not support firmware re-flashing, Zulali used the Arduino to bridge the vehicle's Chrysler SCI serial communication bus with a laptop.
Using Claude for diagnostics and Codex for memory management and coding, Zulali was able to read real-time engine data and eventually implement custom code via an EEPROM chip to adjust fuel delivery parameters. The project, which cost less than $15 in electronic components, has been uploaded to GitHub as an open-source project named xj-scanner.
Entities
Arduino · Chrysler · Claude · Jeep · Muhammad Zulali