The SEO Office Engine™ — run a 25-specialist SEO agency on your own Mac

1 · the idea

What SEO Office actually is

SEO Office is an open-source project by Agrici Daniel (the developer behind the claude-seo skill library). It turns SEO work into an agency you can walk around in.

The whole thing runs on your machine at localhost:3000. When you open it, you're looking at a 3D office floor. Twenty-five desks, one specialist per desk — a technical auditor, a keyword researcher, a schema validator, a backlink analyst, a content strategist, and twenty more.

Click a desk and that specialist goes to work. Or click the orchestrator dais in the middle of the room and tell it what you want in plain English — "build the brain", "run a technical audit", "plan next week's content" — and it dispatches the right specialists in parallel.

Three things make it different from every SEO SaaS you've tried:

It's local-first. Every audit, note and keyword export lives in a folder on your disk. There's no cloud account, no vendor holding your client data.
Each client gets a second brain. Everything the specialists learn becomes a node in a per-client knowledge graph — click any node and read the underlying report.
It's genuinely open source. Licensed AGPL-3.0 — you clone it straight from the author's GitHub and run it yourself. This guide walks through exactly that.

You brief the orchestrator → it dispatches 25 specialists → everything lands in the client brain

2 · under the hood

What happens when you run a sweep

Ask the orchestrator to "build the brain" for a client and the office moves through five phases. Here's the journey.

Intake → diagnostic → discovery → synthesis → one final plan, with gates between phases

Intake sets up the client

The manifest keeper records the client's identity, site URL and business type. A vault linter checks the brain's health before anything heavier runs.

Diagnostic specialists audit the site

Crawlability, indexability, Core Web Vitals, schema markup, sitemaps, on-page issues, international setup. Each specialist writes its findings as a report you can open and read.

Discovery specialists gather the data

Keyword research with volume and difficulty, Search Console queries, GA4 traffic, backlink profiles, competitor pages, image SEO, local signals — pulled through your own API keys.

Synthesis specialists turn data into strategy

Topic clusters, brand positioning, AI-search optimization, programmatic SEO architecture — built from everything the earlier phases learned.

One final plan ties it together

The last specialist reads the whole brain and writes a single prioritised action plan. Every recommendation traces back to a specialist's evidence file — you can check its work.

The Task Feed shows you everything live

Every specialist reports a status — running, review, done, skipped, blocked or failed — colour-coded so you always know whether something's broken or just waiting on a missing key.

3 · the official links

Where to find SEO Office

SEO Office is built and maintained by Agrici Daniel. Everything in this guide comes from his repo and docs — full credit to him:

SEO Office resources ↓

the sourceSEO Office on GitHub → licenseAGPL-3.0 license → watchOfficial demo video → the skillsclaude-seo skill library → dataDataForSEO (pay-per-use) →

4 · before and after

What changes when you run your own SEO office

HOW MOST PEOPLE DO SEO $200–500/mo in tools

Semrush or Ahrefs subscription for the data
A quarterly audit, done manually over days
Findings scattered across docs, tabs and screenshots
Client knowledge lives in someone's head
Every report assembled by hand
Your client data sits in someone else's cloud

WITH THE SEO OFFICE ENGINE™ $0 software · pay-per-use data

Open-source software, running on your machine
A full 25-specialist sweep completes in minutes
Every finding is a node in the client's brain graph
Client knowledge lives in a folder you own
The final plan writes itself from the evidence
Nothing leaves your disk except API calls you chose

The software is free — you pay per-use for data (DataForSEO) only when specialists need it

5 · before you start

Three doubts you might have

Fair questions. Here's the honest picture before you spend 20 minutes installing.

"An AI audit can't match a real SEO consultant."

It's not trying to replace your judgment — it replaces the grunt work. The specialists gather the evidence, run the checks, and write the reports; every claim traces to a source file you can open. You still decide what matters. The difference is a deep audit now takes minutes instead of days, so you can do it weekly instead of quarterly.

"This will need expensive API subscriptions to be useful."

The software itself is free and the LLM side can run on a Claude subscription you already have (it supports the Claude CLI as a provider — no per-token API bill). The data side is pay-per-use: DataForSEO has no monthly minimum, and Search Console and GA4 are free. Missing keys don't break the office — they just switch off the specialists that need them.

"Another tool means another place my client data lives."

This is the one tool where that's backwards. Everything lives in a folder on your disk — audits, keywords, the whole client brain. There's no vendor cloud and no account. The only things that leave your machine are the API calls you configured yourself.

6 · the system

The SEO Office Engine™: four layers

The setup you're about to build has four parts. Each one does a specific job.

The Office

A 3D floor with 25 specialist desks. Click a desk to see that specialist's conversations, files and reports.

ii.

The Orchestrator

The project manager. Brief it in plain English and it dispatches the right specialists in parallel.

iii.

The Brain

A per-client knowledge graph on your disk. Every finding becomes a node you can click and read.

iv.

The Data

Your own keys — DataForSEO, Search Console, GA4, PageSpeed. Pay-per-use, no subscriptions required.

7 · setup

Step 1: Clone and install it

SEO Office is open source — you get it straight from the author's GitHub and it sets itself up with one script.

git clone https://github.com/AgriciDaniel/seo-os.git ~/seo-office
bash ~/seo-office/scripts/install.sh

The installer checks your machine, installs Node 24 and pnpm if they're missing, fetches the dependencies, and copies the example config into place. It tells you exactly what it's doing at each step.

You'll need Python 3.11+ on the machine too (most Macs already have it) — the specialist scripts use it.

"I'm not technical — is this going to be a fight?"

It's two commands, and the install script does the environment setup for you. If anything's missing it prints the exact fix for your machine rather than a cryptic error.

8 · setup

Step 2: Pick your AI provider and keys

SEO Office drives its specialists through an LLM provider you choose. If you already pay for Claude, this part costs nothing extra.

The LLM provider

Open ~/seo-office/.env.local. If you have the Claude CLI installed and signed in, one line uses your existing subscription — no per-token API bill:

SEO_OFFICE_LLM_PROVIDER=claude-cli

It also supports the Codex CLI, the Gemini CLI, or a raw Anthropic API key. Left unset, it auto-picks the first provider it finds installed and signed in.

The data keys (all optional)

# SERP, keyword volume, backlinks — pay-per-use, no monthly minimum
DATAFORSEO_LOGIN=your@email.com
DATAFORSEO_PASSWORD=your-password

# PageSpeed Insights + field data — free Google API key
GOOGLE_API_KEY=your-key

Search Console and GA4 connect through a Google sign-in on the setup page rather than a key in this file. And here's the nice part: a missing key never breaks anything. The specialists that need it just show as gated in the Task Feed, and the rest of the office keeps working.

"What does DataForSEO actually cost?"

It's pay-per-use with no monthly plan — you top up a balance and each SERP or keyword lookup costs fractions of a cent. A full discovery sweep on one client typically costs less than a coffee. Compare that with a fixed $200+/mo tool subscription you pay whether you use it or not.

9 · setup

Step 3: Start it and add your first client

One command starts the office. The first run walks you through setup in the browser.

cd ~/seo-office && pnpm dev

Open http://localhost:3000. On a fresh install it takes you to the setup page — confirm your provider, then add your first client with their site URL.

Then you're standing in the office. Click the spotlit dais in the middle of the room, and tell the orchestrator:

build the brain

Watch the Task Feed light up as the specialists move through intake, diagnostics and discovery. When the sweep finishes you'll have a health score, a stack of evidence reports, and a client brain you can walk around in.

Keep it running (optional)

I registered mine as a background service so it starts with the Mac and stays up. On macOS that's a small launch agent that runs pnpm dev in the ~/seo-office folder — set it once and forget it.

"How long until I see something useful?"

The first sweep on a client site completes in minutes, and the health score plus the prioritised plan land at the end of it. It's the kind of output you'd normally wait days for from a manual audit chain.

10 · setup

Step 4: Give it a tab in the Agent OS

I run SEO Office inside my Agent OS dashboard, next to the rest of the SEO stack — research, OpenSEO, content generation and deploys.

The Agent OS SEO section has a SEO Office tab that embeds your locally-running copy in an iframe, with a live running/not-running badge and a one-click "Open full" button. Since SEO Office serves on localhost:3000, the tab is just a window onto the app you installed in step 1 — the Agent OS doesn't ship or contain SEO Office itself, it points at your own installation.

That's the pattern I use for every external tool in the OS: the tool runs standalone on its own port, and the dashboard gives it a tab with a health check. If the tool's down, the tab shows the exact command to start it again.

The payoff is one screen for the whole SEO workflow: keyword research from Search Console data, OpenSEO for rank tracking, SEO Office for the deep client work, and the content pipeline to ship what the specialists recommend.

"Do I need the Agent OS for this to work?"

No — SEO Office is fully standalone in your browser at localhost:3000. The Agent OS tab is a convenience for people already running the dashboard, so the SEO tools live on one screen instead of five browser tabs.

The SEO Office Engine™ — running inside

Want the whole SEO stack, not just one tool?

This guide covers SEO Office. Inside the AI Profit Boardroom, it's one tab in a complete system — keyword research wired to your real Search Console data, a content pipeline that writes and deploys, and people on live calls who run this on actual client work.

The complete Agent OS — every AI tool you use in one dashboard

SEO pipeline included — research → generate → deploy → index

Free local + free-model engines — run everyday work at $0

Agent Kanban — agents that keep working while you sleep

5 live coaching calls a week — get unblocked on your exact setup

1,000+ prebuilt agents — ready to run and customise

Real playbooks — the exact workflows behind these guides

3,900+ members in 38 countries — someone's online when you're stuck

Join the AI Profit Boardroom → The full system, pre-wired

11 · what members are doing

The results people are getting

The Boardroom is 3,900+ founders and operators — agency owners, ecom founders, course creators — building AI systems into their real businesses. So far, 258 wins have been documented across 38 countries.

3,900+ members inside AIPB
258 documented wins
38 countries
400K YouTube subscribers
163K X followers
29K+ Udemy students

Members post their wins as they happen — SEO results, cost savings, first agents shipped, client work automated. They're all collected in one doc you can read right now.

Read the member wins doc (158 pages) →

12 · the recap

What you end up with

An agency on your Mac

25 specialists that audit, research and plan — dispatched in parallel, done in minutes.

ii.

A brain per client

Every finding in a knowledge graph on your own disk. Nothing lives in a vendor's cloud.

iii.

$0 software cost

Open-source software plus a Claude subscription you may already have. Data is pay-per-use.

iv.

One SEO screen

A tab in the Agent OS next to research, rank tracking and the content pipeline.

One more thing

This is one piece of a bigger system

SEO Office handles the client analysis. The rest of the Agent OS handles everything around it — the keyword research that decides what to write, the content pipeline that writes and ships it, and the agents that keep working after you close the laptop.

If you'd rather build it with help than figure it out alone, that's what the Boardroom is for. 3,900+ people are already inside, and with five live calls a week you're never stuck for long.

Join the AI Profit Boardroom → skool.com/ai-profit-lab

258 documented member wins · 38 countries · 5 live calls a week

Before you go →

Join 3,900+ founders building with this stack.

The AI Profit Boardroom is where the actual Agent OS lives — the templates, the prompts, the daily rooms, the weekly walkthroughs. Same builds you read about here, taught hands-on inside.

3,900+Members

258Documented wins

38Countries

Join AI Profit Boardroom →

No card on this page. Opens in a new tab.