Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

OpenViking

Volcengine's context database for agents, with a Jev rerank client that scores candidate documents with jev-latest as relevance.

Libraryproduction
View on GitHub ↗Live demo ↗

Overview

OpenViking, from Volcengine, is an open-source context database for AI agents that organises knowledge, memory and skills as a virtual filesystem under viking:// URIs. Agents navigate it with ls, tree, read, write and grep, scope semantic search to a subtree, read generated directory summaries before opening full content, and commit sessions as inspectable Markdown memories. A browser studio and self-hosted web studio are available.

Per the list summary, it ships a Jev rerank client that scores each candidate document with jev-latest against api.typesafe.ai and treats the returned probability as relevance, because TypeSafe has no native rerank endpoint. The fetched README excerpt does not cover this client, so the question wording is inferred. It suits teams building agent memory and retrieval who want an optional decision-model reranker.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this document relevant to the query?(inferred)yes · no

How it works

  1. Retrieve candidate documents from the context database.
  2. Ask Jev a relevance question per candidate.
  3. Use the probability as the rerank score.
  4. Return top documents to the agent.

Strengths

  • Rerank option without a dedicated rerank model.
  • Inspectable filesystem-style agent memory.

Caveats

  • Jev rerank client known only from the list summary.
  • One Jev call per candidate document.

At a glance

Type
Library
Status
production
Language
Python
Runtime
database service, web studio
Has code
yes

Use cases and integrations

How curated lists describe it

  • Reranking: Volcengine's agent context database ships a Jev rerank client that scores each candidate document with `jev-latest` against `api.typesafe.ai` and treats the returned probability as relevance, because TypeSafe exposes no native rerank endpoint.

    yibie/awesome-jev

Frequently asked questions

How does OpenViking use Jev?

Through a rerank client that uses Jev's probability as each document's relevance score.

Why not a rerank endpoint?

TypeSafe exposes no native rerank endpoint, per the list summary.

What is OpenViking?

A filesystem-style context database for agent knowledge, memory and skills.

Related projects

  • hippo-memory

    Local long-term memory for coding agents with an optional Jev reranker that orders recalled memories.

    ProjectproductionTypeScript
  • kody

    Multi-user MCP 'agents' cloud' on Cloudflare Workers with an optional search stage that reranks candidates with Workers AI Jev scores.

    MCP serverbetaTypeScript
  • MemSearch Jev reranking

    Zilliz's cross-agent Markdown memory with an optional Jev reranker that asks yes/no relevance questions about retrieved chunks.

    LibraryproductionPython
  • jev-recall

    Filters an assistant's memories by relevance, not resemblance: one Jev yes/no per memory in a single request, kept above a threshold.

    ProjectbetaPython
  • dgui-hypermem

    Self-hosted memory MCP server on Cloudflare Workers where a Jev layer judges memory type, salience and durability and re-ranks recall.

    MCP serverprototypeTypeScript
  • Jev-Mem

    Agent memory system where Jev or local Laya controls memory organisation, query routing, retrieval budgets and stopping.

    ProjectresearchPython