OpenViking abandons traditional RAG vector storage and uses a filesystem paradigm instead. It organizes agent context (memories, resources, skills) under viking:// URIs with a three-tier structure:

  • L0 (Abstract): One-sentence summary for quick retrieval
  • L1 (Overview): Core information and usage scenarios
  • L2 (Details): Full original data, loaded on demand

This enables directory recursive retrieval that locks high-score directories first, then refines content exploration. The retrieval trajectory is fully observable, letting users see exactly how context is being accessed.

Key features:

  • Unified context management via virtual filesystem
  • Tiered context loading reduces token consumption
  • Automatic session management extracts long-term memory
  • Visualized retrieval trajectories for debugging

OpenViking is built with Python, Go (AGFS), and C++. It supports Volcengine Doubao, OpenAI, and LiteLLM providers.

OpenViking GitHub →