Solari split-flap departure boards: Reverse Engineering Solari Airport Departure Boards with Claude Code

A developer used Anthropic's Claude Code to reverse engineer vintage Solari airport split-flap departure boards into a web-based simulation at analogflightboard.com, grounding the AI in 1984 service manuals, patents and physical teardowns so the mechanical specifications came from historical data rather than estimation. The author reports Claude Code handled most of the implementation, including the complex SVG geometry for the animated characters, across 132 commits and roughly 17 hours of active sessions, with the developer directing flip timing, camera angles and sound while Claude generated browser tests to catch interface issues. The author attributes the result to breaking the work into manageable segments with a consistent feedback loop rather than a single prompt.

Evidence at a glance Loading source evidence… What was done A project using Claude Code to reverse engineer vintage Solari airport departure boards, resulting in a functional web-based simulation that allows users to disassemble and inspect virtual flap modules based on 1984 service manuals and patents. The author utilized Claude Code to reverse engineer vintage Solari airport departure boards, creating a web-based simulation available at analogflightboard.com https://reddit.com/r/ClaudeAI/comments/1xi3s3l/i used claude code to reverse engineer the old . The project relied on grounding the AI in historical data, specifically 1984 service manuals, patents, and physical teardowns, to provide mechanical specifications rather than estimations. The author reports that Claude Code handled the majority of the implementation, particularly the complex SVG geometry required for the animated characters. The development process involved using an open-source anatomy skill by wheresryan22 to construct the initial flap module, followed by iterative expansion into a full board. The author directed specific details such as flip timing, camera angles, and sound, while Claude generated browser tests to identify interface issues. The project required 132 commits and approximately 17 hours of active Claude Code sessions. The author notes that success depended on breaking the work into manageable segments and maintaining a consistent feedback loop rather than relying on a single prompt. Community Ask a question, share what worked, or follow the next milestone. Your library https://reversedwithai.com/library