Deploy OpenViking

Deploy and Host OpenViking

OpenViking is an open-source context database for AI agents. It provides file-system-like content organization, semantic retrieval, persistent session memory, and skill management through a unified HTTP API and a built-in Web Studio.

About Hosting OpenViking

This template runs the official ghcr.io/volcengine/openviking:latest image as a single service on port 1933 with automatic HTTPS.

• Persistent Storage: A persistent volume is mounted at /app/.openviking, ensuring accounts, memory, and vector indices survive redeployments.
• Ready-to-Use Defaults: Preconfigured with OpenAI settings. Provide your OPENAI_API_KEY at deploy time; the admin key (OPENVIKING_ROOT_API_KEY) is generated automatically.
• Browser-Only Bootstrap: No CLI required. Open Web Studio, save the root key in /studio/settings, and create your first user in /studio/users.

Common Use Cases

• Give AI agents and assistants persistent, searchable memory across sessions
• Build semantic knowledge bases from documents, code, and resources
• Manage agent skills, session states, and workspace files behind a single unified API
• Safely isolate memory and context across different users and workspaces

Dependencies for OpenViking Hosting

• An OpenAI API key for embeddings and vision-language model processing
• Railway persistent storage mounted at /app/.openviking (provisioned automatically)

Why Deploy OpenViking on Railway?

Railway packages the OpenViking service, pre-tuned configuration, public HTTPS domain, and persistent storage into a single one-click deployment. You can get started immediately in Web Studio without manual server configuration, while retaining full access to environment variables, real-time logs, metrics, and scaling controls.

References

• OpenViking Documentation
• Railway Deployment Guide
• OpenViking on GitHub