Pick an option on each card and press Done; your answer goes straight
back to the session. The raiser's pick is marked. "I'm not sure" and "Take it to the engineering bridge" are real answers.
1
In plain terms. Keep or close PR #20742? Clara wrongly treated a dripping faucet as an emergency on 4 of the last 5 nightly test runs, but the old code gets it right every time I test it, so my fix may not be the real cause.
The full question, as raised
Keep or close the faucet-drip emergency fix (PR #20742)? It was opened because the nightly test saw Clara treat a dripping faucet as an emergency on 4 of the last 5 nights at a company with no custom emergency list. The fix adds a 'match what the problem is, not a shared word' sentence to that company's prompt. But when I ran the old code through the same classifier with every input shape I could rebuild (first message, the photo turn with the conversation in view, with and without weather), it already said 'medium' 18 times out of 18 — so the fix is not shown to fix anything. The PR is a draft now.
Why, and what was tried
The session raising this recommends: Close the fix; reopen once the real cause is found It changes how Clara pages the on-call team at every company with no custom emergency list, and I cannot show it changes the outcome — the old code already got the drip right every time I tested it. A prompt change with no measured benefit is pure risk. The eval improvements (reading the conversation window; four guard cases) are worth a small separate PR. Tried first, unsuccessfully: Measured instead of guessing: ran the real intake classifier with the prompt block swapped to main's on 4 input shapes — 18/18 medium; ran the full 62-case emergency eval suite on the branch (121/124, the 3 misses pre-existing on main); read the nightly logs 10-13..10-17 and the turn-2 snapshot (the alert chain fired on the photo turn, after a token 429 storm). Could not reproduce the red, so keep-vs-close needs a human.