AI diagnostic workbench for electronics repair: upload a schematic and boardview, and the agent traces the fault to the exact pad.
Wrench Board is a diagnostic workbench for independent electronics technicians. You drop in a schematic and a boardview, describe the symptoms, and the agent builds a unified electrical graph, reasons over it, points at the exact pad to probe, reads your measurements, and updates its hypotheses until it diagnoses the fault. The site's pitch: "Repair more boards. Guess less."
Five days, multi-agent Claude CodeAlexis Chapellier — a self-taught builder from a repair shop background in France — prototyped the app in Claude Design, wrote specs and plans for each module, then executed in Claude Code's multi-agent mode, running five or six agents in parallel while debugging. The project took second place in Anthropic's official Built with Opus 4.7 Claude Code hackathon.
By a repairer, for repairersThe big bet was on the model's ability to read visual schematics — Alexis knew it was working when it traced a power path across a motherboard. The project has 244 GitHub stars and is in public beta with a growing community of electronics repairers.