volcengine / OpenViking

### The Context Database for AI Agents
English / 中文 / 日本語
 Website · Live Demo · GitHub · Issues · Docs · Blog 
  Lark ·
  WeChat ·
  Discord ·
  X 
***
## What is OpenViking
OpenViking is an open-source context database for AI agents — one filesystem for everything an agent knows: knowledge, memory, and skills.
Most agent memory is a black box: text goes in, embeddings come out, and nobody can see what was actually stored. OpenViking organizes context as a virtual filesystem under `viking://` instead. Agents navigate it like files — `ls`, `tree`, `read`, `write`, `grep` — and you can open any directory to inspect and edit what your agent knows. Every directory carries a generated summary, so agents can scan summaries first and decide what to read.
Try OpenViking Studio in your browser, no installation required. Self-host Web Studio.
## Why OpenViking
- **One filesystem for knowledge, memory, and skills.** Resources hold documents and code; memories retain user preferences and experience; skills define how to perform tasks — not just extracted facts, but the full context, each with a `viking://` URI for browsing and retrieval. → Viking URI · Context types
- **Search a directory, not the whole index.** Scope semantic search to a project or memory subtree instead of scanning a flat vector pool. `find` runs a query directly; `search` plans retrieval from session context. → Retrieval
- **Read the summary before the source.** Generated directory abstracts (L0) and overviews (L1) let agents judge relevance before opening full content (L2). → Context layers
- **Sessions become files you can read.** Committing a session archives the conversation and extracts memories as Markdown you can inspect, edit, and merge. With VikingBot enabled, `ov compile` organizes source material into a wiki, knowledge graph, or report. → Sessions · Context compilation
Architecture · Design rationale
 [code] 
The three loading tiers:
- **L0 (Abstract)**: a one-sentence summary for quick relevance checks.
- **L1 (Overview)**: core information and usage scenarios for planning.
- **L2 (Details)**: the full original data, read only when needed.
Semantically processed directories carry L0/L1 summaries, so agents can judge relevance before reading full files:
 [code] 
## Proof it works
OpenViking 0.3.22 has been evaluated on long-conversation user memory (LoCoMo) and multi-turn agent tasks (tau2-bench). Full results and setup details, including knowledge-base QA, are in the benchmark report; reproduction scripts live in ./benchmark.
The memory evaluation used Doubao 2.0 Pro as the VLM and Doubao-embedding-vision-251215 as the embedding model.
- **User memory (LoCoMo)**: with OpenViking, all three agent integrations land at 80–83% accuracy — up from 24–57% on their native memory — while input tokens drop by 34.3–91.0% and query latency by 58.45–66.10%.
- **Agent experience (tau2-bench)**: experience memory lifts task success by +6.87pp (retail) and +11.87pp (airline) over the same LL