Act as a senior engineer who has shipped many {{tech_stack}} apps. Everything works on my machine, but on the live site this happens: {{error_message}} Where I see it: {{steps_to_reproduce}} Gather evidence before changing anything: - The exact error in the browser console, and the failing request in the network tab: status, URL and response body - The host's build log, and the server or function logs at the moment it fails - Whether a production build run locally (build, then preview or start) fails the same way. If it does, we can debug locally Likely causes for this symptom, most common first. Check each against the evidence: 1. Environment variables missing or different on the host, or added after the last build. Values the browser needs are baked in at build time and need the framework's public prefix; secrets must never have it 2. Build time versus runtime: code that reads window, localStorage or the current date during server rendering or prerendering 3. File name case: imports that work on a case-insensitive Mac or Windows disk but fail on Linux 4. Paths: a subfolder deploy, hard-coded localhost URLs, or API calls still pointing at the dev server 5. A different database: production keys point at a project missing the latest migrations, policies or seed data 6. Auth settings: the live domain missing from allowed redirect URLs or allowed origins 7. Node or dependency versions that differ between my machine and the build image Confirm the root cause with one targeted check before fixing. Then make the smallest fix and add a check that would have caught it: a startup check that fails loudly when a required variable is missing, a case-sensitive import lint rule, or a CI step that runs the production build. Explain in plain words what was different. Never expose a secret to the browser to make it work.