Scrubby

Quickstart

Get Scrubby running with your codebase in under 5 minutes.

This walks through the fastest path from “no Scrubby” to “your AI editor is using codebase intelligence and your PRs are getting reviewed.”

Prerequisites

1. Sign in

Go to scrubby.ai/dashboard/login and sign in with GitHub. Scrubby uses your GitHub identity to figure out which organizations and repositories you have access to.

2. Choose a plan

Pick a plan from the plans page. New accounts get a 14-day free trial with full access. See Plans & Pricing for the differences between Pro, Team, and Enterprise.

3. Connect your editor

Add the Scrubby MCP server to your editor’s configuration. The exact snippet differs per editor — the Claude Code example:

{
  "mcpServers": {
    "scrubby": {
      "type": "http",
      "url": "https://mcp.scrubby.ai/mcp"
    }
  }
}

Drop that into your project’s .mcp.json. Your editor handles authentication via OAuth on the first connection.

For editor-specific instructions, see:

4. Index your repository

Ask your AI editor to index your repo:

"Index my repo with Scrubby — it's owner/repo-name."

This calls the scrubby_index tool. Indexing typically takes 1–3 minutes depending on repo size. You’ll get an email when it finishes. After the first index, Scrubby tracks new commits incrementally.

5. Start coding

Scrubby’s MCP server tells your editor how to use it via the protocol’s instructions hook. Your editor will consult Scrubby before editing files, before answering architecture questions, and before committing — no extra prompting needed.

For automated PR reviews, install the Scrubby GitHub App on your repositories. It runs independently of your editor integration and provides:

  • Automatic PR analysis on every pull request.
  • Co-change detection (“files you may have forgotten”).
  • Inline code suggestions you can commit directly from GitHub.
  • Convention and consistency checks.

Next steps

Last updated

© Scrubby
Join the beta