Major changes:
- Replace message-count-based memory window with token-budget-based compression
- Add max_tokens_input, compression_start_ratio, compression_target_ratio config
- Implement _maybe_compress_history() that triggers based on prompt token usage
- Use _build_compressed_history_view() to provide compressed history to LLM
- Refactor MemoryStore.consolidate() -> consolidate_chunk() for chunk-based compression
- Remove last_consolidated from Session, use _compressed_until metadata instead
- Add background compression scheduling to avoid blocking message processing
Key improvements:
- Compression now based on actual token usage, not arbitrary message counts
- Better handling of long conversations with large context windows
- Non-destructive compression: old messages remain in session, but excluded from prompt
- Automatic compression when history exceeds configured token thresholds
Improve cron job execution context to ensure proper message delivery and
session history recording.
Changes:
- Add [绯荤粺瀹氭椂浠诲姟] prefix to cron reminder messages to clearly mark
them as system-driven, not user queries
- Use user role for cron reminder messages (required by some LLM APIs)
- Properly handle MessageTool to avoid duplicate message delivery
- Correctly save turn history with proper skip count
- Ensure Runtime Context is included in the message list
This ensures that:
1. Cron jobs execute with proper context
2. Messages are correctly delivered to users
3. Session history accurately records cron job interactions
4. The LLM understands these are system-driven reminders, not user queries
VITOHJL