Bot users
Accounts registered with the Pentest2Bot Telegram bot.
Pentest2Bot ships products that turn an AI agent into an operator: a real browser for web work, a tuning engineer for the machine it runs on, and a scanner that watches the websites you own.
OpenCode · MCP clients · Your server · Your machine
Optimize this machine and check my website
Audit the host: network, disk, security, privacy
✓Apply fixes with approval and rollback
✓Scan the website you own and list findings
✓Schedule a monthly recheck and alerts
✓Counted once per day per product from anonymous install ids. No cookies, no tracking pixels, no personal data.
Accounts registered with the Pentest2Bot Telegram bot.
Machines running Pentest2Browser.
Machines running Pentest2Booster.
Owners waiting for the website scanner.
Daily and weekly activity per product is available at /api/counts.json.
Pentest2Bot is the parent: automation and security tooling for AI agents and website owners. Three products are listed today; more will be added to the same hub.
The parent platform and community. Tools, integrations and research that bring automation and authorized security work together. The bot is also where paid scans and extra targets are managed.
Open the Telegram bot ↗ · — users
Self-hosted browser infrastructure for AI agents. Persistent Chromium profiles, authenticated proxies, compact ARIA observations and replayable workflows through MCP, CLI and REST.
Product page ↗ · GitHub ↗ · — installs
Agent-executed PC and server tuning. One link to your agent and it audits performance, network, security and privacy, applies fixes with approval and verifies every change. macOS, Windows, Server and RDP.
Product page ↗ · GitHub ↗ · — installs
A Chrome extension for website owners. Open your site, run a single check, and get a plain-language report: known vulnerabilities, attack surface, leaked credentials and databases, then automatic monthly rechecks with alerts.
One website free. Extra targets and deeper scans are paid through the bot.
Join the waitlist ↗ · — waiting
Each product exposes one connector or one link. The agent keeps its own model and workflow; the product provides capabilities it cannot fake with text alone.
Add the MCP connector for the browser, paste the booster bootstrap link to your agent, or install the extension from the Chrome Web Store.
The agent works through real interfaces: browser pages, system APIs, package managers, service state. The extension is only the interface — scanning runs on our servers.
Every action produces evidence: before/after metrics, screenshots, findings tables, rollback paths and dated scan reports.
Fetch https://raw.githubusercontent.com/pentest2bot/pentest2booster/main/examples/BOOTSTRAP.md and follow it exactly.curl -fsSL https://pentest2bot.ink/install.sh | sh
# then add the MCP connector to your agentOpenCode is the recommended agent runtime. All products work with any client that supports MCP or shell execution.
If your task lives in a web interface, start with Pentest2Browser. If you want the machine faster, safer or quieter, start with Pentest2Booster. If you own a website and want to know what is exposed, join the Pentest2Extension waitlist.
No. Keep OpenCode, Cursor or any MCP-capable client. You add a connector or a bootstrap link, nothing else.
Nowhere unnecessary. The browser runs on your server, the booster runs on your machine, and counters store only an anonymous daily id. No telemetry payloads and no personal data.
Each install keeps a random local id and sends at most one request per day. That lets us count devices per product without identifying anyone. You can opt out with an environment variable.
Yes. The hub is built to hold additional products; live today are Pentest2Browser and Pentest2Booster, with Pentest2Extension in development.
Explore the products, connect a tool and let the agent do the work.