Course Claude Code Hermes Agent
Hermes Agent MASTERCLASS (Operate & Sell)
Hermes Agent
The Curriculum · 9 parts · 66 lessons · 6+ hours
01

Hermes Foundations

8 lessons · what Hermes is, why it matters — the self-evolving agent
1.1Agent, not a chatbot
  • A chatbot tells you how to book the flight — an agent books it
  • Takes real actions in the real world
  • The one line that changes everything
1.2What Hermes is & stands for
  • Your always-on AI operator that works for you
  • Lives with you · gets smarter · never starts from zero
  • The whole premise in one minute
1.3The self-evolving agent
  • Learns you, Dreams, rewrites its own memory
  • Gets better while you sleep
  • Why it compounds over weeks, not minutes
1.4The anatomy of Hermes
  • The 4 layers: brain · body · memory · the loop
  • What's actually running when it “thinks”
  • The mental model that unlocks everything else
1.5One brain, every interface
  • Telegram · WhatsApp · Discord · Slack · desktop
  • The operator in the middle routing to your agent
  • Same intelligence, every mouth
1.6soul.md — the character Bible
  • Identity, values, voice, standing rules
  • The file at the heart of your agent
  • The template (yours to keep)
1.7Claude Code ↔ Hermes
  • The contractor + the operator — when to use which
  • One shared brain across both
  • Wire the bridge so they think together
1.8Why this is your unfair advantage
  • Capabilities 99% don’t know exist
  • Who it’s really for
  • Where the leverage actually lives
02

Set Up Hermes

7 lessons · get it functional — local, VPS, Docker, always-on
2.1Pick where it lives
  • Your Mac (free) · a VPS · a Docker sandbox
  • The trade-offs of each, honestly
  • Why “always-on” changes what it can do
2.2Install & first run
  • Get Hermes talking on Telegram in minutes
  • BotFather + allowed user IDs
  • Your first real conversation
2.3OAuth vs API key
  • How Hermes authenticates to the world
  • Secure key entry via env vars — never in chat
  • The mistake that leaks your credentials
2.4Run it on a VPS (always-on)
  • Survives reboots & your laptop sleeping
  • 2026 host picks: Hostinger · Hetzner · Railway
  • ~$30–75/mo for a 24/7 operator
2.5Docker, if you want it
  • Containerize for clean, repeatable installs
  • restart: unless-stopped — the survival switch
  • Named volumes so memory persists
2.6Schedule & trigger it
  • cron vs systemd timers · the 6am Dream brief
  • Webhooks: email · Slack · Stripe
  • Let the world wake your agent up
2.7Back it up
  • GitHub backup via cron
  • Don’t lose your memory — off-box backups
  • Recovery in minutes, not hours
03

Powering Hermes

8 lessons · the model masterclass — cloud, local, private & wired to your apps
3.1Own the agent, rent the brain
  • Model-agnostic by design
  • Buying a Lambo and keeping it at 10mph
  • Swap engines without rebuilding the car
3.2The routing decision tree
  • Powerful brain to plan, cheap model to execute
  • Opus/Fable for design · cheap for grunt work
  • The /route thesis, mapped to real tasks
3.3OpenRouter masterclass
  • :nitro :floor :auto explained
  • BYOK, fallbacks, zero-completion safety
  • One key, every model
3.4100% private in your AI lab
  • Run it fully local & free with Ollama
  • The 64K-context catch (Qwen 3 Coder)
  • Vault · Connected · Cloud routing modes
3.5Nous Hermes & open models
  • 70B / 405B, native tool-calling, valid JSON
  • When a local model is genuinely enough
  • The honest gap to frontier (and closing)
3.6The Pantheon
  • Named personas, each with a pinned model
  • Don’t make Einstein wash the car
  • Orpheus · Scribe · Athena — yours to keep
3.7The Triad / Council
  • Plan → Execute → Critique in a loop
  • Opus conductor + cheap worker + critic
  • Never ship anything un-critiqued
3.8Connect it to anything
  • MCP in plain English — one port, every app
  • Gmail & Workspace: read, search, draft — never send
  • Connector security: least access, no rug-pulls
04

Give It a Memory That Never Forgets

7 lessons · the memory core, Graphify & a KB that improves itself
Memory graph · 3D The living memory map from my Claude Code OS
live · always learning recall
Memory core Workspace Vector index Sessions Skills Stale
4.1The 3-tier memory stack
  • Working · session · long-term
  • Why a million-token window is NOT memory
  • The architecture every agent needs
4.2memory.md + recall
  • How Hermes actually remembers you
  • Markdown + full-text search under the hood
  • The “2 weeks to feel magical” curve
4.3Wire external memory
  • Pinecone vectors · Obsidian vault · Supabase
  • Vector vs graph — when each wins
  • Real queries across all sources on camera
4.4The self-improving LLM Wiki
  • Karpathy-style, bidirectional knowledge
  • Rewrites itself, fact-checks, flags contradictions
  • Hermes knows you; the wiki knows your world
4.5Graphify on a schedule
  • Auto-ingest from Granola, email, notes
  • Cron-driven re-index — keep the KB fresh
  • A knowledge graph that updates itself
4.6Memory hygiene
  • Stale facts, decay & TTL policies
  • Conflict detection: merge, update, flag
  • Why retrieval — not generation — is the bottleneck
4.7Memory poisoning (security)
  • The attack nobody warns you about
  • How a hostile email rewrites its mind
  • Write gates + periodic in-context audits
05

Hermes Power Features

6 lessons · goals, Mission Control, Dream, subagents & the commands
5.1The core command set
  • /q /background /kanban /steer
  • /reset /compress /stop
  • The keys you’ll actually use daily
5.2Goals & Mission Control
  • /goal — measurable · scoped · self-served
  • The Super Goal + real-world handshakes
  • AI sprints vs human-only steps, tracked
5.3The Dream engine
  • Overnight self-improvement while you sleep
  • 8 buckets → 4 cards → one morning brief
  • An agent thinking about you, 24/7
5.4Make it proactive
  • From reactive tool to a partner that pings you
  • Goals → Dreams → proactive nudges
  • The triggers that make it feel alive
5.5Subagents & orchestration
  • Spawn workers with isolated context
  • Parallel work without clashes
  • Lead plans → workers explore → lead synthesizes
5.6CLI agent loops
  • printingpress.dev — 100x fewer tokens than MCP
  • The curated loop: brain orchestrates, CLI executes
  • Why agents were trained on the shell
06

Hermes Operating System

7 lessons · the read-only operator console — your AI command center
6.1What the OS is
  • One console for your entire AI stack
  • Read-only by design — safe to watch
  • “Claude Code + Hermes OS”, in one place
6.2The dashboard tour
  • Home · Memory · Skills · Activity · Agents
  • Spend & ROI tracking built in
  • Everything in one pane of glass
6.3The living memory & Intelligence view
  • The 3D memory graph, on screen
  • The knowledge graph & model-intel leaderboard
  • Watch your second brain think
6.4Dream & Mission Control, visualized
  • Your morning brief, in the OS
  • Goals 1:1 synced with Claude Code
  • See what’s the agent’s job vs yours
6.5Install & run the OS
  • Claude-Code-core + Hermes-tier
  • Up and running in minutes
  • Graphify-bridged from day one
6.6The Claude Code ↔ Hermes bridge
  • One shared brain across coding & life
  • Fix the “no handshake” problem
  • Everything connected
6.7Make it yours
  • Pantheon personas & pinned models
  • The model-intel leaderboard
  • Customize the console to your stack
07

Security & Compliance

7 lessons · guardrails, secrets & privacy — keep a real business safe
7.1Risk tiering
  • Read · reversible · external · irreversible
  • Gate by consequence, not the model’s confidence
  • The decision lives in code, not the prompt
7.2The always-gate list & Guardian
  • Deploy · comms · money · delete · permissions
  • PreToolUse hooks: block, modify, auto-approve
  • Why “the agent decides when to ask” fails
7.3Approval gates & kill switches
  • Propose → confirm for anything consequential
  • The 30-minute window so nothing blocks forever
  • How to pull it back mid-run
7.4Spend & loop protection
  • max_turns + token budget hard-stops
  • Circuit breakers & loop detection
  • How agents quietly burn $700 overnight
7.5Secrets management
  • env vars · gitignore and dockerignore
  • System keychain · Doppler · Infisical
  • A key in git history is compromised forever
7.6Observability & cost governance
  • Langfuse / Helicone run-level tracing
  • Console spend caps + per-model budgets
  • Find and kill runaway token burn
7.7Privacy & the law
  • GDPR erasure via per-user memory scoping
  • HIPAA / two-party consent for capture
  • What regulated clients actually need
08

Make Money With Hermes

9 lessons · income on tap — every way to monetize Hermes, for you and for clients
8.1Set up Hermes for others
  • Done-for-you installs — charge to stand it up
  • Turn your new skill into a service
  • The fastest first dollar
8.2The AI back-office retainer
  • Monthly recurring revenue
  • Outreach · reports · scheduling on autopilot
  • The always-on stuff, without babysitting
8.3Sell the outcome, not the hours
  • Value & outcome-based pricing
  • $10–35K builds + $1.5–4K/mo retainers
  • Charge for the operating burden
8.4Land local businesses
  • Roofers · clinics · trades · agencies
  • Niche by vertical — sell fewer wrong assumptions
  • Lead with an ROI calculator, not features
8.5Productize & white-label
  • Package one system, sell it again & again
  • White-label the OS to other operators
  • Recurring beats one-off, every time
8.6Scale your OWN business with it
  • The meta-play: grow what you already run
  • Hermes as your unfair operating leverage
  • Reinvest the hours you get back
8.7Content, audience & sponsorships
  • Monetize building in public
  • Sponsorships · affiliates · a paid community
  • Turn an audience into pipeline
8.8Sovereignty at scale
  • Spin up governed client OSs, repeatably
  • A fleet you run — and resell
  • Multi-tenant isolation · per-client memory
8.9Your daily money-making cockpit
  • The OS surfaces today's highest-value moves
  • Dream prescriptions → revenue actions
  • Proactivity, turned into income
09

The Claude Code Course

your build-side companion · master the craft of building & selling · open the course →
9.1Build real websites
  • Zero to deployed
  • Competitor intel → design → host
  • Spec-first, every time
9.2Build full apps
  • Auth · Stripe payments · Resend email
  • Databases & integrations
  • The BLAST framework
9.3Power features
  • Skills · Hooks · MCP · subagents
  • The autonomous critic loop
  • Dispatch & scheduled tasks
9.4Design systems
  • Claude Code Design · HTML decks
  • 3D · video · AI images
  • Auto-fetch brand assets
9.5Memory + the Hermes Agent
  • Pinecone · NotebookLM · Obsidian RAG
  • Build your own AI operator
  • Wire it all to memory
9.6Ship it safely
  • Compliance · security · monitoring
  • HIPAA / GDPR / API hygiene
  • The Anthropic SDK
9.7Build & sell
  • $30K–$50K deals, zero outbound
  • Pricing & proposals that close
  • The Build & Sell system
AI Automations with Jack Private community · 3,100+ builders Unlock Now All access to my Claude Code and Hermes operating systems.
Everything you unlock
  • FULL Claude Code + Hermes OS
  • The Claude Code Masterclass
  • This Hermes Masterclass
  • 110+ battle-tested blueprints
  • $20K+ in software discounts
  • Daily live calls
  • Personalised 1:1 onboarding
  • 2,500+ AI entrepreneurs
  • Direct feedback & support from Jack
  • 100% tech-support guarantee