Clearfront8 min read

How to Install and Run Clearfront With Claude Code: A Beginner Guide

By Scott, Clearfront founder

You can install Clearfront and run it without writing a single command yourself, by letting Claude Code do the work. Claude Code is Anthropic's AI assistant that lives in your terminal: you tell it what you want in plain English and it runs the commands for you. This guide walks a complete beginner through installing Claude Code, having it set up Clearfront, and then running Clearfront four ways: the interactive REPL, one-shot CLI lookups, the browser console, and as an MCP tool server wired straight into Claude Code so you can scan your footprint just by chatting.

What is Claude Code, and why use it to run Clearfront?

Claude Code is a command-line assistant from Anthropic that reads plain English, then writes and runs the terminal commands for you. Clearfront is a free, open-source tool that scans your own digital footprint. Pairing them means you never have to memorise a command: Claude Code installs Clearfront, launches it, and, through Clearfront's MCP server, can even run scans as a built-in tool.

Two quick clarifications so nothing trips you up. Clearfront has its own built-in AI security analyst, which is separate from Claude Code. And it runs entirely on your own machine, so whatever you scan stays with you. Claude Code is just the helper that gets everything set up and running.

What do you need before you start?

  • -A Mac, Linux, or Windows computer with a terminal. Claude Code guides you through the rest
  • -Python 3.10 or newer, which Clearfront needs to run
  • -A Claude account to sign in to Claude Code, such as Claude Pro, Max, Team, or a Console account
  • -An Anthropic API key for Clearfront's own analyst, from console.anthropic.com, or a local Ollama model if you would rather run it offline

Step 1: Install Claude Code

Open your terminal, run the official installer, then start Claude Code and sign in when it prompts you. On macOS, Linux, or WSL:

Install Claude Code
curl -fsSL https://claude.ai/install.sh | bash
claude

On Windows PowerShell the installer is irm https://claude.ai/install.ps1 | iex. The first time you run claude it opens your browser to log in. After that you are talking to Claude Code: type requests in plain English and it does the rest.

Step 2: Let Claude Code install Clearfront

You do not need to know the install commands. Inside Claude Code, just ask, for example: "Install Clearfront from PyPI with the web console, and set my Anthropic API key." Claude Code shows you each command and asks before running it. Under the hood it runs:

What Claude Code runs for you
pip install "clearfront[web]"
export ANTHROPIC_API_KEY=sk-ant-...

That installs the clearfront command plus the browser console, and points the analyst at your key. If you prefer to run everything offline, ask Claude Code to configure a local Ollama model instead of the key. If you would rather do the manual install yourself, the step-by-step getting started guide covers it.

How do you install Clearfront from GitHub instead?

The pip install above is the quickest path, but you can also install straight from the source on GitHub. Clone the repository if you want to read or change the code or run the very latest version, and fork it if you want to contribute your changes back. Clearfront is MIT licensed, so you are free to do either.

To clone and install, ask Claude Code to run these for you, or run them yourself:

Clone and install
git clone https://github.com/scottmartinanderson/clearfront.git
cd clearfront
pip install -e ".[web]"

To fork it, click Fork on the Clearfront repository, then clone your own copy the same way. If you have the GitHub CLI, Claude Code can do both in one step with gh repo fork scottmartinanderson/clearfront --clone. Either way, set your ANTHROPIC_API_KEY afterwards and every command in the rest of this guide works exactly the same.

The four ways to run Clearfront

Clearfront gives you four interfaces onto the same engine. You can ask Claude Code to launch any of them, or run the commands yourself. Here is what each one is for.

1. Interactive REPL: ask in plain English

The REPL is the friendliest way in. Start it, then type a target or a question. Ask it to check an email, a username, or a domain, and it runs the tools, follows the leads, and writes the report.

REPL
clearfront

Once it is running, type something like "check the email jane@example.com" or "look up the username johndoe99". To get sharper results, see how to prompt the Clearfront analyst.

2. CLI: one-shot lookups for quick checks

When you just want a single answer, or want to script checks and automate them, the one-shot commands run a lookup and print the result with no interactive session.

CLI
clearfront email jane@example.com
clearfront username johndoe99
clearfront ip 8.8.8.8

3. Web console: the chat UI and live evidence graph

The web console is the full chat interface in your browser, with the live, force-directed evidence graph you can click through. Start it and it opens at localhost:8080.

Web console
clearfront web

This is the nicest way to explore results, because you can click a node in the graph and see how one finding connects to the next.

4. MCP: plug Clearfront into Claude Code as a tool

This is the tightest integration. Register Clearfront as an MCP tool server and Claude Code can run scans directly, so you investigate your footprint just by chatting to Claude Code. Add it with one command:

Add Clearfront to Claude Code
claude mcp add clearfront -- clearfront-mcp

The double dash separates the Claude Code options from the command that launches the server. Check it registered with claude mcp list, or type /mcp inside a session. Then you can just tell Claude Code to run Clearfront on an email or username and summarise the exposure. To use it in Claude Desktop instead, add a clearfront entry with the command clearfront-mcp to your MCP config.

Which way should a beginner start with?

Start with the REPL. It is the most forgiving and it explains itself as it goes. Reach for the others as your needs grow.

  • -REPL: your first scans, and any time you want to explore and ask follow-up questions
  • -CLI: quick one-off checks, or automating scans in a script
  • -Web console: when you want to see the evidence graph and click through the connections
  • -MCP: when you use Claude Code daily and want Clearfront on hand as a tool you can just ask for

Whichever you pick, the smartest first move is to run it on yourself. Clearfront scans your own accounts, breaches, and public records in one sweep, on your own machine, and the analyst writes up what is exposed and what to fix first. Install Clearfront free and let Claude Code run your first scan on your own email, username, or domain, or get the free removal guide to start clearing whatever it turns up.

Frequently asked questions

Do I need to know terminal commands to use Clearfront?
No. Claude Code reads plain English and runs the commands for you, showing you each one and asking before it runs. You describe what you want, like installing Clearfront or scanning an email, and it handles the typing.
Is Clearfront's AI analyst the same as Claude Code?
No. Clearfront has its own built-in AI security analyst, which can run on Anthropic Claude or a local Ollama model. Claude Code is a separate terminal assistant that installs and drives Clearfront. The MCP server is what connects the two.
Do I need an Anthropic API key?
Clearfront's analyst needs an Anthropic API key from console.anthropic.com, set as ANTHROPIC_API_KEY. That is separate from how you sign in to Claude Code. If you would rather not use a key, point Clearfront at a local Ollama model and it runs offline.
Can I use Clearfront in Claude Desktop as well?
Yes. Add a clearfront entry to your Claude Desktop MCP config with the command clearfront-mcp, and Clearfront becomes a tool you can call from Claude Desktop, the same way the MCP server works in Claude Code.

I believe your personal data is yours to own and protect. I built Clearfront, a free, open-source tool for scanning and scrubbing your own digital footprint from public data, and I write here about OSINT, breach exposure, and personal privacy.

Scott

Clearfront founder