← Claude Code
Jack Roberts
Fable 5 Agentic OS is Insane... just watch
⏱ 19 min video · 3 min read6 Jul 2026Worth watching
TL;DR
Jack Roberts demonstrates how to integrate Claude 4 Opus (referred to as 'Fable 5' throughout) into a Claude Code agentic operating system across five levels: model integration, unified memory, Ministry of Agents orchestration, autonomous overnight dreaming, and voice interface. The video is a hands-on walkthrough of building a personal AI OS that connects Claude, Hermes Agent, Obsidian, and multiple LLMs into one continuous system.
Key points
1
Level 1: Add Claude Opus 4 (Fable 5) to the Hermes Agent model selector with adjustable effort dials (low/medium/high/extra high) so it can analyze your entire agentic OS and surface unused high-ROI skills
2
Level 2: One Brain unified memory connects Claude Code chats, Hermes chats, and an Obsidian local wiki so all AI tools share the same context lake — Claude Opus 4 can then restructure this into a clickable, searchable, topic-clustered knowledge explorer
3
Level 3: Ministry of Agents places Claude Opus 4 as the orchestrator over sub-agents (DeepSeek V4 Pro, GLM 5.2, GPT-5.5) using prompt caching via OpenRouter to dramatically cut token costs while getting multi-model consensus answers
4
Level 4: Dreaming functionality lets Claude Opus 4 autonomously review all your data overnight — Claude Code activity, memory files, chat logs, automations — and return structured improvement suggestions in the morning
5
Level 5: A voice interface (Voice Line) built directly into the Claude Code OS dashboard lets you verbally query your entire operating system, check stats, and even trigger live UI changes like color updates
Actionable insights
→
Use a low effort/token setting for broad discovery tasks (e.g. skill gap analysis) and reserve high effort for actual build tasks to avoid burning credits unnecessarily
→
Connect your Obsidian vault to both Claude Code and Hermes Agent so all models draw from the same local wiki, eliminating repeated context-setting across tools
→
Set Claude Opus 4 as the orchestrator in the Ministry of Agents setup and route cheaper models (DeepSeek V4 Pro, GLM 5.2, GPT-5.5 via OpenRouter) as sub-agents to get high-quality consensus answers at a fraction of the cost
Notable quotes
“If AI had infinite context about you, what your business goals were, what your financial numbers were, what you spoke to about yesterday and a year ago — you would never need to repeat yourself and you would always get the best insights from it.”
“Some say Jack it's mixture of agents. I'm saying hey, Ministry is way cooler.”
“Having a super intelligent AI to review all of your stuff while you are sleeping, working for you, I think is an incredible benefit.”
Worth watching?
✅
Worth watching the full video?
Worth watching if you are actively building with Claude Code and Hermes Agent and want a live walkthrough of wiring them into a multi-agent OS — but you can get most of the architecture and prompts from this summary alone.
Saved you some time? The creator still deserves a like.
Found this by searching? That's the hard way.
summree watches the AI channels you follow and sends you the summary the moment a video drops. Nothing gets past you. From £4/month.
Try it free