Block a user
Ask up to three times, and say so when she gives up (review of PR 1)
Add the pending-question data layer for clarifying questions
Fixed in PR 23. All five points, plus two traps found on the way out (#382, #383) and one mute path that had nothing to do with your review — finishClarified could return an empty reply and say…
Fix the phrasing prompt: she was reading the placeholder aloud (0/15 to 13/15)
Add a voice.llm_router flag (default off) and fix kill-maven.sh
Not quite — and the confusion is my fault for not saying what the script does at the top.
kill-maven.sh only kills. It never picks up a running server and never spawns one. There's no…
Add the pending-question data layer for clarifying questions
You're right, and five of your six comments are the same bug wearing different hats. I'm fixing it — replies in order.
**1. MaxAttempts should be 3 or 5, and she should say so when she gives…
Model bake-off: keep Qwen3.5-0.8B, LFM2.5 is worse in Russian
Gate note recall on the margin, not the raw score
Route with the resident model by default
Run slot extraction on LLM router decisions
Let the router say it does not know
Record the quiet-box router numbers
Salvage two eval harnesses: nudge phrasing and per-model routing
Make make test really check gofmt and vet
Make accepted routines keep firing instead of nudging once
Fix the router prompt: questions were filed as facts (48.7% to 60.5%)
Stop the full nudge body leaving the box, and survive a panicking sink
Measure how often Maven finds the right note
Make snooze actually suppress nudges
Ask one clarifying question, then finish the request