Shane 938629df83 feat: LLM-based context compaction with idle trigger
When conversation exceeds 24 items (~12 turns), Gemma summarizes the older
messages into a 2-4 sentence recap that replaces them, keeping the last 10
items verbatim. Also triggers after 5 minutes of idle time so returning
users get a compacted context rather than a bloated one.

Falls back to hard truncation if the summary call fails.
2026-08-22 16:51:21 -04:00
2026-08-22 16:26:01 -04:00
S
Description
Real-time voice assistant: LiveKit + Azure Speech + local Gemma LLM
426 KiB
Languages
Python 59.3%
JavaScript 21.8%
CSS 11.7%
HTML 3.8%
Dockerfile 2.4%
Other 1%