One Claude skill that stands up your entire agency operating system: the engine, the memory, your Google suite, and skill packs, on a single dashboard. It replaces three weekends of wiring. Setup takes about ten minutes.
You run an AI agency. A client asks for lead routing, so you build lead routing. Next month they want a reporting flow, so you build that too. Each job ships and then sits on its own, one more thing you maintain by hand.
None of it compounds. You end the quarter with a drawer full of one-off automations and eight browser tabs open to run any of them.
The agencies pulling ahead sell something else. They sell the operating system that produces the automations. Same Claude work under the hood, packaged as one product a client installs and keeps.
What the OS actually is
Five parts, wired together so they act as one thing.
Claude Code is the engine. It runs the business tasks. A memory layer sits underneath it so the OS remembers what it did and who it did it for. Your Google account is connected, so Gmail, Calendar, and Drive are real surfaces the OS can read and act on, not screenshots you paste in. Skill packs for research, content, ops, and data come switched on. And one dashboard sits on top: a command center that shows what the OS can do, what is scheduled, and the latest results from a single screen.
That is the thing you use yourself. It is also the thing you package and sell to a client, or hand to someone on your team so they can run it without you in the room.
Building that from scratch used to eat three weekends. This skill does it in one run.
Copy the Agentic OS Installer
Paste this into Claude Code as a skill, or drop it into a new chat and say "follow this skill." Fill in the four fields at the top first.
Copy this.
--- name: agentic-os-installer description: Ship a complete Claude Code powered Agentic OS for your agency instead of building it over three weekends. This skill installs the whole operating system that produces your automations: Claude Code as the engine, a memory layer, your Google suite wired in, and skill packs for research, content, ops, and data, all rolled into one dashboard command center. Use it when you run an AI agency and want to stop selling one-off automations and start installing the system that produces them. --- # Agentic OS Installer > What this does, in one line: you tell Claude about your agency stack and it > stands up your whole Agentic OS (engine, memory, Google suite, skill packs, > and one dashboard) so you have a working operating system instead of a pile > of disconnected automations. ## Before you start: fill these in These are the only things you provide. Everything else is automatic. | Fill in | What it is | Example | |---------------|-------------------------------------------------------------------|-----------------------------| | [AGENCY_NAME] | What to call this OS so it is branded to you | Bright Studio OS | | [VAULT_PATH] | The folder on your machine where the OS and its memory will live | /Users/you/BrightStudioOS | | [GOOGLE_ACCOUNT] | The Google account whose Gmail, Calendar, and Drive to connect | you@brightstudio.com | | [SKILL_PACKS] | Which capability areas to switch on first | research, content, ops, data| If you do not have one of these yet, the skill tells you exactly where to get it the first time you run it. Start with one Google account and add more later. ## What you need installed - Claude Code (the engine the OS runs on). Codex works too. - A note app for the memory layer. Obsidian is the simplest, pointed at [VAULT_PATH]. - A Google account you are willing to connect for Gmail, Calendar, and Drive. - That is it. The skill scaffolds the rest. ## How to run it 1. Put this file where your agent looks for skills (in Claude Code, a skills/agentic-os-installer/ folder), or paste this whole file into a new chat and say "follow this skill." 2. Fill in the fields above. 3. Tell your agent: "Install my Agentic OS at [VAULT_PATH] for [AGENCY_NAME]." ## The workflow (what the agent does) 1. Confirm the inputs. If [VAULT_PATH] does not exist, create it. This folder is both the OS and its memory. 2. Scaffold the memory layer: a clean vault with folders for the OS itself, clients, notes, and a daily log. This is where every run gets remembered. 3. Wire in the Google suite for [GOOGLE_ACCOUNT]: connect Gmail, Calendar, and Drive so the OS can read and act on real work. Walk the user through the one-time sign-in if it is not connected yet. 4. Drop in the requested [SKILL_PACKS]. For each area (research, content, ops, data), install a small set of plug-and-play skills the agency can run on day one. 5. Build the command center: one dashboard that surfaces what the OS can do, what is scheduled, and the latest results, all from one screen. 6. Brand it to [AGENCY_NAME] and write a short README so a new team member or a client can pick it up without you in the room. 7. Hand back the working OS plus a one-line summary of what was installed and what to run first. ## If something goes wrong - Google will not connect: sign in with [GOOGLE_ACCOUNT] and approve Gmail, Calendar, and Drive scopes. Re-run the connect step after approving. - A skill pack does not show up: check the pack folder landed under your skills directory inside [VAULT_PATH], then restart your agent so it re-reads skills. - The memory layer feels empty: that is expected on day one. It fills in as the OS does work. Ask it to log every run to the daily note. - You want to sell it to a client: point the installer at a fresh [VAULT_PATH] and a different [GOOGLE_ACCOUNT]. Each install is its own clean OS to hand over.
How to use it
Save the file where your agent reads skills, or paste it straight into a fresh chat. Fill in the four fields at the top so the install is branded to you and points at the right folder. Then tell Claude to install your OS at the vault path, and let it work through the scaffold.
What you get back
01. The engine. Claude Code set up as the thing that runs your business tasks, not one automation at a time but as the standing worker behind all of them.
02. The memory. A vault with folders for the OS, your clients, your notes, and a daily log. Every run is written down, so the OS remembers the last client call instead of starting cold each morning.
03. Your Google suite, connected. Gmail, Calendar, and Drive wired in for the account you name. The OS reads real threads and files and acts on them.
04. Skill packs, live. Research, content, ops, and data, each a small set of ready skills you can run from the first day. Add your own on top whenever you want.
05. One dashboard. A command center that shows what the OS can do, what is scheduled, and the latest results. This is the screen you use, the product you sell, and the handoff you give your team.
Honest: where this stops
The skill scaffolds the OS. It does not read your mind. The skill packs are starting sets, not a finished agency in a box, and you extend them as your work gets specific.
The memory is empty on day one. That is correct, not broken. It fills as the OS does real work, so the first week looks quieter than the second.
Each client install is its own vault and its own Google account. Do not run two agencies through one OS. Point the installer at a fresh folder and a fresh account per client, and every deployment stays clean and ready to hand over.
And it does not lock you to one model. The OS runs on Claude Code or Codex, so the operating system is yours even if you swap the engine underneath it later.
Do one thing this week
Point the installer at an empty folder and run it once. Watch it stand up the vault, connect a single Google account, and drop in one skill pack. Do not wire the whole agency yet. Just see the OS assemble itself in front of you.
If you do not like what it builds, you spent ten minutes. If you do, you have the first version of the thing your best competitors are already selling.