Paste the stack trace, error log, or screenshot of the failure. Include whatever context you have — the deploy that triggered it, the request payload, the environment. Grok reads the full trace, identifies the failing component, and explains the error in plain language.
Example prompt
We're seeing intermittent 502s in production after deploying commit abc1234. Here's the Nginx error log and the upstream Node.js stack trace. It only happens under load — roughly 5% of requests to /api/checkout.
Grok traces the failure to its origin — whether it's a connection pool exhaustion, a race condition, a missing null check, or a config mismatch. It proposes a targeted fix with the exact code change, and explains why it works.
Example prompt
The pool exhaustion theory makes sense — we increased the checkout concurrency limit in that deploy but didn't bump the Postgres max connections. Show me the config change and the code fix. Also, how should we set up connection pool monitoring so we catch this earlier next time?