Integration

Scrubby + VS Code

VS Code ships with first-class Model Context Protocol support. GitHub Copilot's agent mode and any other MCP-aware AI extension can pick up Scrubby out of the box. Once connected, the AI agents you already use in VS Code get architectural awareness of your repo — domains, conventions, co-change patterns — with no custom extension to install.

What you get

Codebase-aware Copilot agent mode. When Copilot is in agent mode, it can call Scrubby's tools directly, the same way it calls read_file or run_in_terminal.
Workspace-scoped configuration. Drop the MCP config in your repo's .vscode/mcp.json and every contributor who opens the workspace gets Scrubby for free.
Pre-commit changeset checks. Ask the agent to run scrubby_review_changeset on staged files before pushing.
Lower token usage. The agent stops grepping aimlessly. One Scrubby call returns a structured answer.

Setup

1. Add the MCP server

Create .vscode/mcp.json at the root of your workspace (VS Code reads this file automatically; you can also put the same config in your user profile's mcp.json):

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

VS Code provides IntelliSense for this file, so it'll autocomplete and validate as you type.

2. Authenticate

The first time an MCP-aware extension connects to Scrubby, you'll be prompted to authenticate via GitHub OAuth in a browser tab. Sign in with the GitHub account you used for Scrubby.

3. Confirm the server is running

Open the Command Palette (Cmd/Ctrl + Shift + P) and run MCP: List Servers. Scrubby should appear with its tools listed and a green status indicator.

4. Index your repository

From your AI agent's chat (Copilot agent mode or whichever extension you're using), ask it to index:

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

Indexing takes 1–3 minutes. After that it stays current incrementally.

Tools available to VS Code agents

Tool	Purpose
`scrubby_index`	Index a repository for the first time, or re-index after major changes
`scrubby_review`	Get domain, conventions, connected files, and git history for a specific file
`scrubby_review_changeset`	Check changed files for missing co-changes and convention violations
`scrubby_get_domains`	List all architectural domains discovered in the repository
`scrubby_get_segments`	List code segments with their conventions and file summaries
`scrubby_get_network`	Get the domain connection graph showing how areas of the codebase relate
`scrubby_get_findings`	Retrieve findings from previous reviews
`scrubby_report_findings`	Report findings so Scrubby can refine its connection weights

Notes for VS Code specifically

Workspace vs user config

Workspace config (.vscode/mcp.json) is the right place for project-specific servers like Scrubby — checking it into git means every contributor who clones the repo gets it. Use the user profile's mcp.json only for MCP servers you want available everywhere.

Copilot agent mode vs chat mode

MCP tools surface in agent mode, where Copilot can autonomously call tools. In plain chat mode, you'll need to prompt for Scrubby calls explicitly or switch the agent on.

Sandboxing

Scrubby is a remote HTTP MCP server, so VS Code's stdio sandboxing settings don't apply. Auth and traffic flow over HTTPS to scrubby.ai.

Use any MCP-aware extension

Scrubby works with anything in VS Code that speaks MCP — GitHub Copilot's agent mode, Continue, Cline, and others. The same .vscode/mcp.json entry serves all of them.

Troubleshooting

Server not showing in MCP: List Servers — check that .vscode/mcp.json exists at the workspace root and reload the window (Developer: Reload Window).
Tools not exposed to Copilot — make sure you're in agent mode, not the default chat mode.
OAuth never completes — check your default browser is reachable and not blocking the popup. You can also visit your Scrubby dashboard and reset the session.

Bring codebase intelligence into VS Code.

Join the Scrubby beta Read the setup docs →