Published: 2026-05-16

Agent OS: Run Claude, Hermes & OpenClaw Together From One Dashboard

Chapters / key moments (click to jump — plays here on the page)

Julian Goldie demonstrates a locally-hosted "Agent OS" — a Next.js mission control dashboard he built in a single Claude Desktop session without writing any code. The system unifies Claude, Hermes, and OpenClaw into one interface with shared memory: all agent conversations are automatically logged to an Obsidian vault, and that vault is fed back to every agent as context, so each one knows your goals, projects, and history. The dashboard also tracks session analytics, skill inventories, API keys, and Kanban tasks synced from agent chats.

Source video

"Agent OS: Claude + Hermes AI = Superpowers!" by Julian Goldie SEO — Watch on YouTube →

Key Takeaways

The three-layer model: Claude as the brain layer, Hermes as the execution layer, OpenClaw as the connection layer — all managed from one locally-hosted dashboard.
Build it without coding: prompt Claude Desktop to "create a mission control dashboard for managing my AI agents," specify a Claude CLI bridge, and iterate until the UI is what you want.
Shared memory via Obsidian: every agent interaction auto-saves to an Obsidian vault; the vault is fed back to all agents so they share context across sessions and platforms.
Omi (a wearable recorder) captures daily screen and microphone activity, generating 1,200+ personal memory entries that get indexed into Obsidian for agent context.
Dashboard includes per-agent analytics (sessions, tool calls, token usage, peak hours), a goal tracker, skill management, and session history — all in one browser tab.

What you can actually set up from this

Extracted from the video's own transcript — the specifics the original summary left out.

Configuration shown

Build prompt used in Claude Desktop (as shown)

Create a beautiful operating system hosted locally for managing Claude ... It should be like a beautiful mission control dashboard. Then allow me to control my OpenClaw, my Hermes, any other agents in separate systems inside the dashboard.

Template shown in the video. Adapt to your project — do not copy verbatim without reviewing each line.

Reproducible steps

Answer the build questions
When Claude asked how the dashboard should talk to Claude, he chose a Claude CLI bridge. He accepted the recommended Next.js + Tailwind stack and gave it the Hermes and OpenClaw documentation.
Add a memory layer
Every exchange with each agent is auto-saved into an Obsidian vault, and agents read context back from it. A journal and a goals tracker are stored in the vault too.
Iterate by asking for changes
Follow-up requests added voice input, per-agent chat, a control room per agent (API keys, sessions, skills, Kanban, analytics), and export.

Gotchas

Keep a dashboard like this local-only. It holds API keys and a vault of personal notes.
Hermes in the demo ran on a free API model, so the agent layer can cost nothing; building the dashboard used Claude Opus.
Heavy promotion for a paid community; the full dashboard code is not in the video.