• 30 modular tools, email, username (sherlock + WhatsMyName), broad username discovery across 3,400+ sites (maigret), search-based footprint discovery, IP, IP self-exposure report, domain, WHOIS, breach, Gravatar profile, EmailRep reputation, phone, paste, EXIF/GPS metadata, Shodan, VirusTotal, Censys, IP2Location, AbuseIPDB, GitHub (profile + public code/secret exposure), DNS, subdomain discovery via certificate transparency (crt.sh), historical URL recovery via the Wayback Machine (Internet Archive), mass-scan visibility (GreyNoise Community), infostealer-exposure check (Hudson Rock, free tier, no plaintext credentials), dork generation, live dork search, URL scraping, BTC/ETH address lookup, and passive domain recon (theHarvester)
• MCP server built in, expose all 30 tools natively to Claude Code, Claude Desktop, and any MCP-compatible client
• Three AI backends, Anthropic Claude (default), local Ollama, or any OpenAI-compatible endpoint; tool results come from real subprocess calls, never hallucinated
• Fully async, parallel tool execution via asyncio.gather() with hard subprocess timeouts
• MIT licensed, no embedded LLM; bring your own API key or run fully offline

Legal Disclaimer: Clearfront is intended for legal and authorized use only. Users are solely responsible for ensuring their use complies with all applicable laws and regulations. The authors accept no liability for misuse. See DISCLAIMER.md.

What is Clearfront?

Clearfront is an AI agent for Open Source Intelligence with five interfaces: an interactive terminal REPL, a direct CLI, a browser-based web console, an MCP server exposable to Claude Code, Claude Desktop or any MCP-compatible client, and an agent skill for any client following the Agent Skills standard. The AI layer uses Anthropic's native tool use API (or a local Ollama model, or any OpenAI-compatible endpoint): the model issues hard stops when it needs a tool, your code executes the real binary, the actual output goes back, hallucination in tool results is structurally impossible.

Installation

pip install clearfront

Or from source, if you want to modify it:

git clone https://github.com/scottmartinanderson/clearfront
cd clearfront
pip install -e .

External binaries (must be in PATH):

If a binary is absent, the corresponding tool returns a descriptive error string. All other tools remain operational.

Quick Start

# Interactive AI REPL (default)
clearfront

# Web interface
clearfront web

# Direct tool (no AI)
clearfront email target@example.com

Configuration

Store all keys in a .env file at the project root (copy .env.example). python-dotenv loads it automatically at startup.

The Bright Data link above is a referral link; signing up through it supports Clearfront at no extra cost to you.

Optional Python packages:

Tools

Interfaces

Interactive REPL

Run clearfront with no arguments to start the AI-powered REPL. Type a target (email, username, domain, IP, name) or a question; the agent decides which tools to run, chains them on findings, and compiles a report.

REPL commands: <target>, clear, save, tools, config, history, help, exit / Ctrl-D.

All sessions are auto-saved to ~/.clearfront/history/. Browse with clearfront history.

Web UI

pip install "clearfront[web]"
clearfront web
# Opens http://localhost:8080 automatically

Browser-based AI chat with streaming tool output, inline result cards, and a light/dark theme toggle. Supports fully local inference via Ollama or any OpenAI-compatible endpoint (no Anthropic API key required when using a local backend).

The console runs entirely locally and binds to 127.0.0.1 by default. Choose your backend and paste your own key in Settings; your keys and the targets you investigate never touch our servers. Screenshots are in media/.

MCP Server

Expose all 30 tools to any MCP-compatible AI client.

Claude Code:

claude mcp add clearfront python /absolute/path/to/clearfront/mcp_server.py
claude mcp list

Claude Desktop, add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "clearfront": {
      "command": "python",
      "args": ["/absolute/path/to/clearfront/mcp_server.py"]
    }
  }
}

Agent Skill

skills/clearfront-osint teaches an agent how to drive Clearfront: which approach fits a given request, the direct subcommands, how to read the confidence ratings, and the authorized-use rules. It follows the Agent Skills open standard, so it works in Claude Code, Cursor, Codex, Copilot, VS Code, Gemini, Windsurf, Zed and the other skills-compatible clients.

npx skills add scottmartinanderson/clearfront

Install Clearfront itself as well (pip install clearfront); the skill is the instructions, not the tool. Use the skill when you want the agent to decide how to run a sweep, and the MCP server above when you want it calling the 30 tools directly.

Docker

docker compose up --build
docker compose run --rm clearfront email target@example.com --json

Set ANTHROPIC_API_KEY (and optionally HIBP_API_KEY, IPINFO_TOKEN) in a .env file or export them before running. Reports persist to ./reports/ via a volume mount.

CLI Reference

Contributing

Issues and pull requests are welcome. See CONTRIBUTING.md for the development workflow and coding conventions. Please read DISCLAIMER.md before contributing.

License

Clearfront is open source under the MIT License.

The bundled username dataset clearfront/tools/data/wmn-data-unique.json is a filtered adaptation of the WhatsMyName project by Micah Hoffman, used under the CC BY-SA 4.0 license; that file (and adaptations of it) remains under CC BY-SA 4.0. See clearfront/tools/data/NOTICE.

For authorized security research only. See DISCLAIMER.md.