Get oriented.
The vocabulary and shape of agent-native work. Read these in any order before you build anything.
What Is Claude Code?
Three ways to use Claude on your computer: the CLI, the desktop app, the VS Code extension. When to pick which one.
What Is an Agent-Native OS?
The orientation card. Eight components that make up an agent-native OS, mapped to the workshop stages that build each one.
Your Brain File: CLAUDE.md
The persistent context file your agent reads every session. What goes in it, what doesn't, and how to keep it sharp.
Slash Commands and Skills
Two ways to give your agent a workflow. Slash commands run on your call. Skills auto-trigger when Claude spots the right context.
The /personalize Flow
Twenty-one questions about you, your work, and how you learn. Claude turns the answers into a CLAUDE.md tailored to you.
The Modes Claude Can Run In
Five modes for how much your agent does without asking, from asking before every action to running with no prompts. The autonomy ladder you climb as trust grows, and the two you have to turn on first.
Running Claude Safely
Permissions, backups, and the safety net. Stop worrying about accidentally deleting things. Set up the layers that make trust possible.
Secrets Management
Why your agent's API keys belong in a vault, not a chat or config file. The three paths (Infisical, 1Password, a locked .env) and why Infisical is the fast, safe default. The why before the build.
Upgrade to 8D Secrets
Make your vault live. Install the Infisical Agent and a machine identity so your agent can read vaulted keys headlessly, for scripts, MCPs, and scheduled jobs.
MCP Servers: Doors for Your Agent
How your agent reaches the outside world. Doors to Gmail, Calendar, Notion, Slack, and anything else with an API.
GitHub for Knowledge Work
Commits, pushes, branches, PRs. Plain-English explanations for non-developers who got handed a GitHub account and bounced. The workshop survival guide.
The Agent OS Stack
A picture of every layer in the system. From the chat window down to the secrets vault. Useful when something breaks.
Terminal Basics
Never typed a command into a terminal? Start here. Six commands cover most of what the workshop needs.
Plain English Glossary
A dictionary of every term you'll hear. Plain-speak definitions, no jargon, no homework. Bookmark it.
Connect your tools.
Give your agent its senses. Install your first MCP, borrow a working CLI, and stand up a basic second brain. Everything in the next stage builds on these.
Install Your First MCP
Firecrawl as the cleanest useful first install: one free API key, no OAuth dance, no Google Cloud project. One command and you can verify the doors actually open before you move to Gmail or Calendar.
Borrow a Working CLI
Pre-built CLI tools you can install today: Slack, Zoom, Google Workspace, N8N, Firecrawl, MacWhisper. Inherit the plumbing, write your own skills on top.
gws CLI Connect
Install the Google Workspace CLI, authenticate one Google account, and feed Gmail plus Calendar into your Morning Brief without raw exports in git.
n8n CLI Toolkit
Full command-line control of your n8n instance for you and your agents. Create, debug, and monitor workflows without the UI. 80+ commands, a 543-node catalog, zero dependencies.
Slack CLI Toolkit
Full programmatic control of Slack from the terminal. 306 API methods, raw passthrough, and 33 starter skills, so your agent can post, read, and manage Slack.
Zoom Automation Hub
Programmatic control of Zoom via Server-to-Server OAuth. Schedule meetings, pull recordings and transcripts, and read reports from the terminal.
Know When to Build a CLI
Some services already have great MCPs. Others need a custom CLI. This is the decision tree: when to build, when to use what's shipped.
The Cairns Three Tiers
Memory in three layers: waypoints on top, dense cards in the middle, raw evidence at the base. The architecture every other knowledge build sits on.
Mini Cairns Build
Stand up your own Cairns second-brain vault. Folder shape, daily ritual, and the agent contract that runs it. The hands-on for Component 02.
Second-Brain Capture
One skill that takes anything you throw at it (links, voice memos, screenshots) and files it where it belongs. Raw preserved, card generated.
Ship your first agent.
Pick one. Small builds you can stand up in an afternoon, each wired to the tools and memory you just connected. Every one teaches a piece of the capstone.
iMessage Status Check
Text "what's on my plate today" to your agent. Get a clean summary back in ten seconds. No computer needed.
Meeting Transcript Processor
Granola transcripts become structured notes with attendees, decisions, and action items. Filed automatically.
Calendar Conflict Defender
Meeting requests get auto-scheduled around your deep work. Conflicts come back with a draft decline and a counter-time.
Voice Memo to Journal
Dictate while walking. The agent transcribes, structures, and files the entry to your second brain. Tagged and searchable.
Slack Session Auto-Drafter
Cohort session posts draft themselves from the prior session's transcript and the upcoming agenda.
Email Reply Queue
Drafts replies to client emails older than 48 hours, written in your voice, queued in Gmail for one-click send.
Stale Lead Follow-Up
Opportunities untouched for 14 days get a draft check-in email tailored to the last real conversation.
Go deeper.
The advanced layer: sub-agent teams, custom CLI architecture, and a second brain that grows into real semantic and graph search.
Sub-Agents, Background Agents, and Agent Teams (Oh My!)
One agent isn't always enough. Three primitives plus four named patterns from Hunter's merge-2-main library. Six agents running from a single prompt becomes the new normal.
Build Your First CLI
Four phases: scrape the API docs, build the tool, write per-method skills, cache the full spec for everything else. The Slack CLI walkthrough ships 306 methods this way.
The 4-Layer CLI Architecture
CLI tool, per-method skills, full API catalog cached locally, plus meta-skills that compose across multiple CLIs. The shape behind every CLI shipped in this repo.
Wrangle Many Skills
One CLI ships with one gateway skill that routes to a local model holding the full skill library. Solves the auto-discovery ceiling without bloating context.
The /build Command
One slash command, every blueprint. /build reads any blueprint and runs the same six steps every time. Unblocks the agent-that-builds-agents capstone.
Starter Cairns Packs
Three pre-built knowledge bases that ship with the workshop. Research papers, tool docs, and YouTube. Borrow them as-is or use as scaffolding.
Chain-of-Density Cards
The compression technique behind every Cairns card. Five rewrite passes, each one denser than the last, evidence preserved.
Cairns Ingest Pipeline
Drop in an article, a transcript, a note. The pipeline preserves the raw, generates a dense card, and proposes new waypoints. Every step you approve.
Memory Linting
Audits your vault for weak waypoints, missing cards, broken backlinks, and duplicates. Run weekly to keep memory honest.
Decision Logging
Capture every meaningful decision: reasoning, who agreed, what could reverse it, when to revisit. Filed to your second brain.
The Librarian Agent
A small local agent that knows every memory tier and decides which one to read first. The thing that keeps your main context lean.
Vector Search vs Graph RAG
Two ways agents find information. Search by meaning catches similarity. Knowledge graphs catch relationships and time. Most production builds use both. When to reach for which.
YouTube Cairns Blueprint
Watch 40 channels at once. Every new video gets transcribed, tagged, embedded, and dropped into your knowledge vault overnight.
Make it autonomous.
Schedules, heartbeats, fleet. The "while you sleep" moment when agents start running themselves.
Heartbeats and Crons
Two ways to make agents wake up on their own. Heartbeats poll on an interval. Crons schedule at fixed times. Most builds use both.
Cairns Dream
Sleep mode for your second brain. End-of-day pass distills the day's activity into cards and proposes memory updates. Never auto-writes.
Multi-Machine Heartbeat
Agents on different computers ping each other every five minutes. Anything that needs human attention escalates to Slack.
Langfuse Tracing
Auto-traces every Claude Code session and tool call. Costs, latencies, prompts, responses, full audit trail.
Fleet Observability Dashboard
One board for every agent run across every computer you own. Cost, duration, outcome, and full trace.
N8N + Claude Integration
Three ways to wire Claude Code to n8n workflows: push from Claude, pull from n8n, or scheduled polling. Pick what fits.
Production-grade.
Deployed, observed, on-call. The projects behind real client work and the capstone builds.
Build Your Own Brand Kit
Scaffold an AI-readable design system for yourself or a client. Seven-folder shape: a brand spec, design tokens, and a Claude Skill. The SKILL.md is the AI-native unlock.
The Morning Brief
The workshop capstone. Sources collapse into one daily ping plus a deployed dashboard. Vercel, Supabase, a Telegram or email delivery, and a cron that runs while you sleep. Every other concept in the workshop feeds into this build.
Ship Anything Autonomous
The deployment pattern from the Morning Brief, abstracted. Supabase for state, Vercel for the dashboard, env vars from your password manager, a cron on your Claude subscription. Apply the same moves to any agent shape you want running on a daily loop.
Grow Room Monitor MCP
Polls AC Infinity sensors on a schedule, logs time-series data, renders a self-contained HTML dashboard. Zero SaaS dependencies.
Triple-Model Bid Prep
Three CLIs write the same bid in parallel. A fourth agent triangulates and flags conflicts. Standing operating procedure for real client work.
Agent That Builds Agents
Give your /build skill a one-sentence problem. It returns a deployable workflow. The meta-capstone.
Second-Brain Auto-Router
Every transcript, voice memo, screenshot, and saved article gets routed to the right Obsidian note. Zero manual filing.