# Connect your site

> The last step - one command into your coding agent, and it installs its own pipeline.

Source: https://dispatchseo.com/docs/connect-your-site
All docs in one file: https://dispatchseo.com/llms-full.txt

With the backend running, connecting your site is one paste. This is the
step where DispatchSEO stops being a dashboard and becomes a pipeline.

## What you need

- Somewhere to publish: **self-hosted WordPress** (connected from Settings
  with an application password, articles posted straight to it), or your
  site's source in a **GitHub repo** (articles arrive as pull requests).
  Hosted site builders like Wix and Squarespace have neither and won't work.
- **An AI to drive it** - the ordinary Claude app at claude.ai, or a coding
  agent: Claude Code, Codex, or Cursor. Your AI is the engine; it runs on your
  existing plan or key. The steps below are the coding-agent path. For the
  Claude app, the dashboard's **Connect your AI** screen gives you a connector
  address to paste into claude.ai instead of a command to run, and Home shows
  a "Your next step" list that ticks itself off as you go: connect, let Claude
  ask its three questions, ask for ideas, approve them on the Queue screen,
  ask it to write, watch Drafts. The first time Claude uses each tool,
  claude.ai asks you to allow it - choose Always allow.

## The one command

1. Log in to your dashboard and add your site as a project (the setup
   wizard does this with you; this page exists for adding more sites
   later).
2. The dashboard shows you a single command with your project's key
   already filled in. Copy it.
3. On your own computer, open a terminal **in your website's repo
   folder** (the folder with the `.git` in it) and paste the command. It
   connects your coding agent to DispatchSEO and walks you through the rest,
   explaining each thing it asks for before asking.

Your agent does the rest of the install itself: it connects to the MCP
server, reads the served instructions, writes its own workflow files into
your repo, and sets its own repo secrets. When it's done, your repo has a
content pipeline and the dashboard knows about your site.

## What happens after

1. **Your agent researches.** It mines keywords from your Search Console
   data, Google Autocomplete, and what it already knows about your product.
   Ideas land in the dashboard queue with reasoning attached.
2. **You approve, or don't.** Each idea is a card: the keyword, why it's
   winnable, the angle. Approve, reject, reorder - or flip on auto mode and
   skip the queue.
3. **The pipeline builds.** Every morning a GitHub Action turns the oldest
   approved idea into a real pull request against your repo - a guide or a
   small free tool, checked against the live SERP.
4. **It tracks what happened.** Daily rank checks, hourly Search Console
   snapshots, index verification. When a job breaks, you get a red banner
   and an email instead of silence.

## Multi-site

One deployment manages any number of sites. Each project gets its own MCP
token, its own data, its own settings - add the next site the same way,
from the dashboard.