# OpenViking

OpenViking is a context database for AI agents: one filesystem for what an agent knows (documents and code you import, memories from its conversations, and skills), with search over it. It is open source and self-hosted, installable on Unraid.

Canonical page: https://ourlocalhost.com/app/9019/openviking

## Description

OpenViking is a context database for AI agents: one filesystem for what an agent knows (documents and code you import, memories from its conversations, and skills), with search over it. Agents reach it over an HTTP API, an MCP server, the ov command line, or Python, TypeScript and Go SDKs. It includes a web interface, OpenViking Studio, for browsing and searching that content. Requirements: an account with a model provider that offers an embedding model and a vision-language model (VLM): OpenAI, Volcengine, Kimi, GLM and others, or Ollama for local models. No GPU is needed in this container. First start: the container waits for its settings file, ov.conf. Until it exists, every request returns a message saying so. Open the container's Console from the Docker tab and run: openviking-server init In the wizard: - Choose your provider and models for the embedding model and the VLM. - At

## Platform availability

Available on 1 of 10 platforms: Unraid.

## Health score

79/100 (Good), based on 3 of 4 factors.
- Maintained: 100 (last commit 0 days ago)
- Popular: 93 (42,564 GitHub stars)
- Easy to install: 20 (1 install platform, Docker image)

## Links

- Unraid: https://ca.unraid.net/apps/app_27291b037562f96f
- GitHub: https://github.com/volcengine/OpenViking
- Official website: https://openviking.ai

## Common questions

### Can I run OpenViking's web interface on my own server?

Yes. OpenViking Studio can be used in a browser, and the repository includes a self-host Web Studio option.

### What should I decide before provisioning a self-managed OpenViking server?

Review the deployment checklist and choose who will operate the HTTP service. If you have a VikingDB or OpenViking private deployment package, use Enterprise Deployment and operations.

### How can I inspect the knowledge, memories, or skills stored for an agent?

OpenViking organizes context as a virtual filesystem. You can open any directory to inspect and edit what the agent knows.

### How are conversations kept after an agent session?

Committing a session archives the conversation and extracts memories as Markdown. Those files can be inspected, edited, and merged.

### Can semantic search be limited to a project or memory subtree?

Yes. Semantic search can be scoped to a directory or subtree instead of scanning a flat vector pool. find runs a query directly, and search plans retrieval from session context.

---
Source: Our Local Host (https://ourlocalhost.com/app/9019/openviking), the self-hosted and open-source app directory.
