Blog in Public.
Technical notes from the OpenViking team — on agents, protocols, and the systems behind them.
Deploy a Fully Local OpenViking on NVIDIA DGX Spark with cuVS
Start from a clean DGX Spark: install Ollama local models, OpenViking, and NVIDIA cuVS, configure GPU vector search, run it as a service, and verify the path from import to read-back.
Should You Use OpenViking? A Three-Minute Test for Agent Wranglers
Twelve questions reveal your agent-wrangler type, a five-axis profile and an OpenViking fit score, then tell you whether open source, the managed service or a private deployment fits best. Or hand a prompt to your own AI and let it judge.
Let the Agent Turn the Page: Codex-Style Context Management on Pi, Backed by OpenViking
Codex now lets the model decide when a context window ends. We rebuilt the mechanism on Pi with OpenViking’s existing session archive, and a model with no training for it finished a task larger than one window.
Memory That Follows You: OpenViking × Agent Harness
A demo of memory across Claude Code, Codex, and two Python projects, with a written guide, session recording, and timestamps.
OpenViking Now Supports Agent Plugins 1.0
One portable package brings OpenViking memory to ChatGPT, Codex, Cursor, GitHub Copilot, Kiro, and VS Code — no more per-client plugins.
From Assistant to Colleague: Building Multi-User Agents with OpenViking's Peer Mode
Standard AI memory breaks in group chats. Learn how OpenViking’s Peer Mode decouples identity from context, turning passive bots into true team colleagues.
Agent Swarms Don’t Lack Scale — They Lack Memory
Field notes from Agent Hub, a persistent space shared by humans and heterogeneous agents: hosting memory and identity so a resident stays itself through restarts—and even a brain transplant.
Stop Teaching Every Agent From Scratch
How OpenViking turns frontend design preferences, project decisions, and agent handoffs into shared long-term memory.
OpenViking User / Peer: Separating Data Owners From Interaction Objects
Why OpenViking moved from a User / Agent mental model to User / Peer, and how one agent can serve many people without turning every participant into an OpenViking user.
OpenViking Benchmark Update: User Memory, Agent Memory, and Knowledge Base QA
May 2026 OpenViking benchmark update across LoCoMo user memory, ClawWork and tau2-bench agent memory, and HotpotQA / single-turn RAG knowledge-base QA.
OpenViking for the Too Many Agents Problem
A practical OpenViking deployment and Web Studio walkthrough for sharing context across Claude Code, Codex, Hermes Agent, Manus, Lovable, Cursor, and MCP clients.
VikingBot: When Agent Memory Starts Changing Strategy
A bilingual rewrite of the VikingBot werewolf experiment, showing how OpenViking turns multi-agent chat history into durable, inspectable, strategy-changing memory.
OpenViking for Claude Code & Codex: Give Your Coding Agent Persistent Memory
One command gives Claude Code and Codex long-term memory powered by OpenViking: conversations are captured and distilled, recalled when relevant, and shared across clients.
OpenViking: Inside the Context Database Architecture
How OpenViking turns directory semantics, decoupled storage, identity, performance, and privacy into a context database layer for AI agents.
Building an Agent Daemon
Four runnable demos plus one OpenViking integration step that turn a CLI agent into a managed daemon with message delivery and shared durable context.
OpenViking: The Database Paradigm for Context Engineering
How OpenViking turns context engineering into a database-shaped interface for agents.