Cut only the offending sentence on leak, keep the rest of the reply (audit M9)

Fold Cyrillic homoglyphs in injection prefilter, live probe 09.10.2026

Cap mush-scan length, document cooldown instead of health registry

Fix small hardening issues across logging media and errors

Gate based-on identity leaks on first-person form

Require injection context for debug-god-dan mode matches

Drop unused _normalize_for_detection helper

Close ban guest bypass and quarantine echo poisoning

Re-check injection filter after voice transcript; widen identity-leak and normalize prefilter (A3-1, A3-2, A3-3)

Comment cleanup: short Russian why-comments, no AI boilerplate

Cleanup wave 6: one shared SSE parser and one shared fallback chain, drop dead code and a silent list alias, cover the previously untested failure branches

Comment polish: Russian pro-style comments, keep dated evidence short (+426/-2765). Fix: bot.send_chat_action -> bot.bot.send_chat_action (was AttributeError, typing indicator never showed)

Rich Messages sending, review fixes, no-file note hygiene, persona tone

Hide model internals from users, English internals in logs, Claude-style tone, live-log model updates

Model audit Sept 2026, speed-aware routing, waiting dots, trigger and heuristic polish

Add files via upload

Add files via upload