Context And RTK

Inferoa uses context settings and RTK to reduce token waste while preserving the evidence required for accurate work.

Context Settings

context:
  compression_threshold: 0.8
  context_window: 32768
  protected_recent_loops: 3
  engine:
    provider: auto
    startup: welcome
    require_ready_before_chat: true
    watch: true

compression_threshold controls when Inferoa starts compacting older context. protected_recent_loops keeps recent model and tool work visible. The context engine can use automatic detection, CodeGraph, built-in behavior, or be turned off.

RTK Settings

rtk:
  enabled: true
  delivery: managed
  version: 0.42.3
  auto_download: true

Environment overrides:

INFEROA_RTK=false inferoa
INFEROA_RTK_PATH=/path/to/rtk inferoa
INFEROA_RTK_AUTO_DOWNLOAD=false inferoa

Operational Views

/context
/tokenmaxxing

/context reports context usage and compression state. /tokenmaxxing reports RTK savings and token pressure alongside endpoint usage evidence.