Rivet's lesson: JEV doesn't reason—split severity classification into tiny yes/no prompts
yawnxyz · x · 2026-11-19
Hard-won tip from the Rivet team: JEV does not reason, by design. Their initial single prompt asking for an SEV1–4 severity verdict gave mediocre results; after reading the docs they split it into tiny binary asks (did production go down? are they blocked with no workaround? is something broken?) and mapped answers to severity with if-branches, which worked like a charm. The screenshot shows the decision process: checks run top to bottom until one matches, e.g. 82% confidence on "something in Rivet is broken" → SEV3.
More from coding & agent
- New TLAPS-Bench blog: formally proving correctness of real-world system specs with TLA+ — tianyin_xu · 2026-11-19
- 16x Microsoft MVP demos a native PowerShell agent built on Microsoft's Intelligent Terminal — dfinke · 2026-11-19
- Musk: Treat Grok Bot like a really smart employee, says most users underuse it — elonmusk · 2026-11-19
- Eve agent tool cuts build time 8x to 6.9s by reusing compiler module graph — evilrabbit_ · 2026-11-19
- Dev shares NL-driven testing UI: 'the best interface is chat' — KlausCodes · 2026-11-19
- Devs flag GPT's defensive programming for hiding bugs; OpenAI asks for /feedback cases — pvncher · 2026-11-19