OpenViking

What is OpenViking?

OpenViking is a context database for AI agents, built by the Viking team at Volcengine. Instead of scattering memory, resources and skills across vector stores, code and modules, it puts them in one directory tree addressed by viking:// paths that agents browse with commands like ls, tree, read and grep. Each directory carries a short L0 abstract of up to 256 characters and an L1 overview of up to 4,000 characters, so an agent reads summaries first and opens the full file only when needed. Once a session is committed, memories are pulled out in the background and saved as readable Markdown. It connects to Claude Code, Codex, Cursor, TRAE, OpenCode, OpenClaw, Hermes and any MCP client. You can self-host the open-source edition under AGPLv3 for free, use the managed OpenViking Context service on Volcengine with the first 50 files free and hourly pay-as-you-go billing after that, or arrange a self-managed deployment in your own VPC.

Use cases

• Give coding agents persistent memory across sessions
• Organize agent knowledge as a navigable file tree
• Share skills and context between multiple agents
• Replace ad hoc vector store memory modules
• Isolate context per user or peer
• Deploy a context store inside a private VPC

Who it's for

• Agent developers
• AI engineering teams
• Platform teams
• Enterprises

Why use OpenViking

• Agents keep context between sessions, so you correct a convention once instead of every day.
• One filesystem interface for memory, knowledge and skills is easier to inspect and edit than opaque vector chunks.
• Summary-first loading helps keep prompts smaller by reading full files only when needed.
• You can start free on your own hardware and move to the managed service with the same kernel.
• Works across many coding agents, so one context store can serve a whole toolchain.

OpenViking features

• viking:// directory paths for memories, resources, sessions and skills
• Tiered L0 and L1 summaries with L2 originals on demand
• Recursive retrieval scoped to a directory
• Background memory extraction stored as Markdown
• ov CLI and MCP access with the same operations
• Memory plugins for Claude Code, Codex, Cursor, TRAE and OpenCode
• Multi-tenancy with account, user and role permissions
• Envelope encryption with your own keys in the open-source edition
• OV Compile and snapshots
• OVPack migration path from self-hosted to cloud
• Docker and Helm deployment

OpenViking integrations

• Claude Code
• Codex
• Cursor
• TRAE
• OpenCode
• OpenClaw
• Hermes Agent
• LangChain
• LangGraph
• Manus
• Ollama
• Volcengine Ark
• Lark
• GitHub

How to get started with OpenViking

1. Choose a server: the hosted OpenViking service, your own self-hosted server, or a new deployment.
2. Run the installer, which detects tools such as Claude Code, Codex or Cursor and installs the memory plugin.
3. Provide an API key for the hosted service, or point the plugin at your own server.
4. Restart your agent so it can write memories and resources into viking:// paths.
5. Memories are extracted in the background and retrieved in later sessions through search or directory browsing.

OpenViking limitations

• Memories are processed in the background, so a fact is not instantly recallable after being said
• Each integration depends on a running OpenViking server, whether self-hosted or managed
• The managed service runs in Volcengine's Beijing region and needs a verified Volcengine account
• The open-source edition runs on a single node or replicas you assemble, with no automatic failover
• AGPLv3 licensing may need review if you offer a modified version as a service

OpenViking pricing

OpenViking is open source, so you can use and self-host it for free.

OpenViking plans

• Open Source

  Free

  free forever, AGPLv3

  • Full feature set, self-hosted
  • No feature gates or license keys
  • Community support
  • OVPack migration path to cloud

• OpenViking Context

  First 50 files free

  then pay-as-you-go hourly

  • Fully managed on Volcengine
  • VikingDB vector engine
  • Encryption, audit, network isolation
  • Commercial SLA

• Self-Managed

  Custom

  • Your VPC or air-gapped deployment
  • Distributed deployment with license activation
  • Vendor support and priority fixes

Plans as listed on the official pricing page on 24 Oct 2026. Prices can change and may vary by country or exclude tax. Check the official pricing page.

OpenViking alternatives

The main OpenViking alternatives listed on Debuly are Mem0, Zep, Letta, Supermemory, Pinecone and Chroma. The table compares pricing and platforms; order is A–Z.

1. 

   Chroma

   Open-source search database for AI apps, with vector, full-text, regex and metadata search and a serverless cloud option.

   Developer platformsResearch and searchAutomation and agents
   Open sourceVisit Chroma (official site)
2. 

   Letta

   Platform, formerly MemGPT, for building stateful AI agents with persistent memory, from a terminal app, SDK or API.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Letta (official site)
3. 

   Mem0

   Memory layer for AI agents and apps, available as an open-source package or a managed platform that keeps user context across sessions.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Mem0 (official site)
4. 

   Pinecone

   Managed vector database for AI search and RAG, with dense, sparse and full-text indexes, inference and an Assistant API.

   Developer platformsResearch and search
   FreemiumVisit Pinecone (official site)
5. 

   Supermemory

   Memory and context API for AI agents with user profiles, a memory graph, retrieval and connectors, billed by usage.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Supermemory (official site)
6. 

   Zep

   Managed memory and context platform for AI agents, built on the Graphiti knowledge graph, with Python, TypeScript and Go SDKs.

   Developer platformsAutomation and agents
   FreemiumVisit Zep (official site)

OpenViking FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from res.gcloudcache.com · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.