Powered by Pingu Unchained 4 Abliterated LLM Learn more →
Your AI.
No Limits.
No Objections.
Hire an autonomous AI pentester with unlimited AI tokens that works on its own to keep your targets secure, 24/7, no sleep, minimal supervision.
Point it at a target and it handles the rest — recon, scanning, exploitation, and reporting. No content filters. No refusals. Runs on dedicated GPU infrastructure. Connect via WhatsApp, Telegram, Discord, or API.
WhatsApp Telegram Discord Signal Slack
$20/mo — unlimited AI tokens included.
agent@penclaw:~$ scan --target app.example.com --depth full
[*] Enumerating subdomains... found 14 hosts
[*] Running port scan... 3 open services detected
[!] SQLi vector found: /api/v2/users?id=1' OR 1=1--
[!] XSS reflected: /search?q=<script>alert(1)</script>
✓ Report generated — 2 critical, 1 medium, 4 info findings
Your pentest team, always on
An AI agent that lives in your workspace — scanning, testing, and reporting vulnerabilities around the clock.
🔓
Truly Uncensored
Built on Pingu Unchained 4 Abliterated LLM. No safety theater. No refusals. Real offensive security analysis without artificial limits.
🕵️
Automated Recon & Scanning
Subdomain enumeration, port scanning, service fingerprinting, and vulnerability detection — all driven by conversational commands.
💬
Operate From Anywhere
Run pentests from Signal, Slack, Discord, Telegram, or WhatsApp. Just message your agent — it handles the rest.
⚡
H100 GPU · Unlimited Tokens
Running on NVIDIA H100 with 80GB VRAM. 128K context window. Flash attention. Fast inference at ~195 tokens/sec. No token limits — use as much as you need.
📋
Instant Reports
Get structured vulnerability reports with severity ratings, proof-of-concept payloads, and remediation guidance — delivered right in chat.
🔑
API Access
OpenAI-compatible API endpoint included. Integrate pentesting into your CI/CD pipeline or build custom security workflows.
Unlimited AI usage — included in $20/mo
Unlimited Cloud Code with Kairos Mode
Your PenClaw instance runs 24/7 in the cloud — making it the perfect home for Claude Code's Kairos Mode. An always-on AI coding agent that never sleeps, powered by unlimited Pingu Unchained 4.
Want to use a different LLM? Bring your own API key to swap models anytime. Unlimited Pingu Unchained 4 stays available as your default.
🤖
Personalize with Meta-Claw
Use Meta-Claw to customize your OpenClaude instance — install tools, configure settings, and tailor the agent to your workflow, all conversationally.
Streamed directly to your browser
Your Cloud Desktop
Get a full Linux desktop environment streamed directly to your browser. Run GUI security tools, web browsers, file managers, and anything else you need — all in the cloud.
All included — unlimited OpenClaude with Pingu Unchained 4, Claude Code Kairos Mode, terminal, file editor, desktop GUI, and all security tools. BYOK to swap LLMs.
50+ security MCP servers in registry
MCP & Skills Marketplace
Pre-loaded cybersecurity MCP servers ready to install. Or deploy any MCP repo from GitHub in one click — no rejections, always works.
☁️
ScoutSuite
Cloud Security
🎯
VirusTotal
Threat Intel
Your Cloud Pentesting Stack
🖥️
Cloud Desktop GUI
Full Linux desktop streamed to your browser. Run GUI tools, browsers, and file managers. OpenClaw operates it alongside you.
📝
Cursor IDE in the Cloud
Full coding environment on your own tenant. Edit files, write scripts, manage configs from a browser-based IDE.
🖳
Web Console
Full terminal access to your instance. Run tools, inspect logs, execute commands — all from the browser.
🤖
Meta-Claw Agent
An AI agent with sudo that retrofits your OpenClaw instance. Install packages, deploy MCPs, patch configs — conversationally.
⚡
Unlimited OpenClaude
Unlimited AI coding agent on the Claude Code harness with Pingu Unchained 4. Kairos Mode runs 24/7 on your cloud instance. BYOK to swap LLMs. Personalize with Meta-Claw.
📦
Skills Marketplace
50+ pre-loaded security MCPs. One-click install for hosted servers. One-click deploy for self-host repos via Audn.AI.
How One-Click Deploy Works
1
Find MCP on GitHub
Any MCP server repo
▲
Deploy to Vercel/Render
One click, no config
✓
Paste URL in PenClaw
Instant connection
Not just pentesting — your AI, your rules
Built for pentesting. Ready for anything.
PenClaw ships as a pentesting preset — but your instance is a full cloud AI workspace. Use reset-identity to wipe the pentester persona and build something completely different. Automate workflows, run bots, build agents — it's your machine.
Here's what people have already built:
🚀
Your imagination?
You can obviously do so many things with this. An always-on cloud instance with unlimited AI, computer use, a full desktop, and a Claude Code harness — the possibilities are endless.
Email us or tap the blue chat button on the bottom left and tell us what you're building. We'll list your use case right here.
Simple pricing
One plan. Full access. Unlimited OpenClaude. No hidden fees.
⚡
Single Dedicated GPU
Assigned exclusively to you
Pro
$20 /mo
Everything. Unlimited OpenClaude. Unlimited AI tokens.
- Unlimited AI tokens — no metering
- Pingu Unchained 4 — zero censorship
- Dedicated instance · 128K context window
- Signal, Slack, Discord, Telegram, WhatsApp
- Full configuration dashboard
- OpenAI-compatible API endpoint
- Unlimited OpenClaude on Claude Code harness
- Claude Code Kairos Mode — 24/7 always-on agent
- BYOK — bring your own API key to swap LLMs
$20/mo. Cancel anytime.
$20/mo — unlimited AI tokens included.
FAQ
What is PenClaw?
PenClaw is an AI pentester agent with unlimited AI tokens that lives in your workplace. It runs Pingu Unchained 4 — an abliterated LLM on an NVIDIA H100 80GB GPU. No token metering, no usage caps. Includes unlimited OpenClaude on the Claude Code harness with Pingu Unchained 4. Kairos Mode works out of the box since your instance runs 24/7 in the cloud — personalize it easily with Meta-Claw. Bring your own API key to swap LLMs anytime. You operate it through Signal, Slack, Discord, Telegram, or WhatsApp.
What's free vs. what costs money?
Free: All marketplace skills, hosted MCP servers (Audn Sec QA, Supermemory), and the MCP Server Manager are free to use with your subscription. HOPX Sandbox is managed by HOPX with its own free tier. For self-host MCP servers that need deployment, you get one-click deployment with a 1-day free trial on Render-managed infrastructure. After the trial, Render's standard pricing applies for hosting. Your $20/mo PenClaw subscription covers the AI agent, unlimited tokens, and all built-in skills.
What does "abliterated" mean?
Abliteration removes the safety fine-tuning from a base model. The result is an AI that performs real offensive security analysis without corporate safety filters or refusal patterns getting in the way of legitimate pentesting work.
How do I connect my messaging apps?
After subscribing, your dashboard lets you configure channel connections. Add your Signal, Slack, Discord, Telegram, or WhatsApp credentials. Each channel is isolated to your instance — just message the agent and start pentesting.
Is my data private?
Yes. Each subscriber gets an isolated PenClaw gateway. Your scans, conversations, and findings never mix with other users. The model runs locally on our infrastructure — nothing is sent to third-party APIs.
How do I get the most out of PenClaw?
1. Install Featured MCP Servers — Start with the four featured items in the Marketplace: Audn Sec QA (free, voice & text agent red-teaming), HOPX Sandbox (HOPX managed, cloud code execution), Supermemory (free, persistent memory for the agent), and MCP Server Manager (free, manage all your MCP connections). These are one-click installs and give you the core toolkit.
2. Understand What's Free — All skills in the marketplace are free. Hosted MCP servers (like Audn Sec QA and Supermemory) are free. HOPX Sandbox is managed by HOPX. For self-host MCP servers, you get a 1-day free trial with one-click deployment to Render.
3. Use Supermemory for Persistence — Install Supermemory so your agent remembers previous scans, targets, and findings across sessions. This builds a knowledge base that makes each subsequent pentest smarter.
4. Run Continuous Scans — Point the agent at a target and let it run autonomously. Use the Console to monitor progress, and check Session Logs to review findings. The agent works best when given clear scope and left to enumerate, scan, and report.
5. Pair Recon with Action — After enumerating emails or subdomains, ask the agent to cross-reference with OSINT sources, check for leaked credentials, or test authentication. Chain multiple tools together for deeper coverage.
6. Deploy More MCP Servers — Browse the marketplace for 50+ security MCPs. Deploy any GitHub MCP repo in one click via pingu.audn.ai/mcp.
Can I integrate it into my CI/CD pipeline?
Absolutely. Your instance exposes a standard OpenAI-compatible API. Use it to run automated security scans on every deploy, integrate into GitHub Actions, or build custom pentesting workflows.
Ready to hire your AI pentester?
Deploy an abliterated AI agent with unlimited tokens into your workplace in under 60 seconds. Operate it from any channel.
$20/mo — unlimited AI tokens included.