Dipstick Alerts – Search recalls and service bulletins for your car

Hacker News

Share card

Dipstick Alerts – Search recalls and service bulletins for your car

Hi HN! When I bought my first car, a 2019 Subaru Forester, I also bought an extended warranty. I wanted to stay informed about problems that might be covered, so I started checking NHTSA.gov—the website of the National Highway Traffic Safety Administration, a U.S. agency—every month for new service bulletins. Those bulletins pointed me to problems I might not otherwise have recognized as repairable. For example, my trunk would not open as far in cold weather. I probably would not have asked the dealer about it, but I found a service bulletin describing the problem and was able to get it fixed under warranty. I wanted a simpler way to find information relevant to my car and to be notified when something new appeared, so I built Dipstick Alerts. This was also my first substantial project built with extensive help from AI coding tools. You can use the site without signing up: select a year, make, and model to search for recalls and manufacturer service bulletins. Results include a plain-language summary, symptom notes, suggested next steps, and a link to the original manufacturer document when available. Email alerts are optional. NHTSA is the underlying data source, so coverage is currently focused on vehicles sold in the United States. Some of the information may still be useful elsewhere because many vehicles and components are sold globally, but models and bulletin applicability can differ between markets. A daily process scans for newly published recalls and bulletins, imports them through queues, matches them to applicable vehicles, tracks superseding bulletins, and prepares alerts for subscribers. Some of the plain-language content is generated from the source documents using Gemini. The original manufacturer documents remain the authoritative source. The application runs on Cloudflare Workers using D1, Queues, KV, and email sending. I’d especially appreciate reports of incorrect vehicle matches and feedback on whether the summaries are genuinely useful, whether anything in the search or alert flow is confusing, and how the site performs under load. I’ll be around to answer questions and fix problems. Thanks for trying it!