LOADING

ASTRONOX

Initializing runtime 8%

MintFlow

Capabilities

Desktop-native AI, with stronger judgment layers

Astronox is built around controlled execution, provider orchestration, memory relevance, and practical desktop awareness rather than generic chat polish alone.

Auto routingLocal toolsPrivacy-firstCompanion mode

Provider Orchestration

Blend cloud and local models without manually babysitting every choice

Astronox supports Gemini, OpenAI, Claude, Azure, Ollama, MintAI/OpenRouter, and Auto routing with backoff, fallback, and local-model awareness.

  • Auto mode evaluates complexity, availability, and recent provider health
  • Fast Mode forces lighter, lower-latency routing when speed matters most
  • Pro adds MintAI for managed multi-model access without manual key juggling

Control Surface

Keep risky actions explicit and observable instead of hidden inside magic

Astronox is designed around confirmations, scoped execution, and user-visible judgment instead of opaque automation.

  • Safe and Full access modes for local tooling
  • Explicit confirmations for higher-risk actions
  • Peacock suggestions pause execution until you decide how to proceed

Context Layers

Combine short-term conversation state with durable long-term learning

Astronox uses a layered memory architecture that keeps recall relevant and local by default.

  • Short-term conversation context and mid-term vector recall
  • Trident Memory System for post-session consolidation into memdir files
  • HookWare ingestion and response grading that sharpen personalization over time

Desktop Integration

Work from the machine you already use instead of a browser-shaped abstraction

Astronox exposes OS settings, current media, system search, companion window workflows, and a command surface built around desktop reality.

  • Current song detection, playback control, and OS settings shortcuts
  • Companion window with dock presets and quick model controls
  • Supersearch command palette across actions, settings, and core UI surfaces