PentestAgent: Open-Source AI Framework for Penetration Testing with Attack Playbooks

PentestAgent is an open-source AI agent framework for penetration testing. Developer Masic from GH05TCREW released it on GitHub. It uses LLMs like Claude Sonnet or GPT-5 via LiteLLM for black-box security assessments.

The tool runs in a terminal user interface with three modes. Users pick assisted chat, single autonomous agent, or multi-agent crew setups. Prebuilt attack playbooks handle web app tests like THP3 workflows. Legal warning stresses authorized targets only to avoid law violations.

BEST WINTER DEALS

Editor's Choice

Private Internet Access

Access content across the globe at the highest speed rate.

70% of our readers choose Private Internet Access

70% of our readers choose ExpressVPN

ExpressVPN

Browse the web from multiple devices with industry-standard security protocols.

Nord VPN

Faster dedicated servers for specific actions (currently at summer discounts)

Playbooks launch via CLI commands. They guide AI through recon, scanning, and exploitation steps. RAG systems inject domain knowledge. Notes save to loot/notes.json and build shadow graphs for crew mode insights.

Built-in tools include nmap, sqlmap, Playwright browser, and Tavily web search. TUI controls like /target, /tools, and /report give full command. Docker images pack Kali tools like Metasploit.

New HexStrike integration adds advanced pentesting workflows. MCP servers expose scoring tools. Setup scripts handle dependencies on Windows, Linux, or Mac.

GH05TCREW repo README states: “PentestAgent is an open-source AI agent framework for penetration testing. Use only on authorized systems.” 

HexStrike docs confirm: “Model Context Protocol framework for pentesting tools integration.”

Core Modes Table

Setup Process

Clone repo and run platform scripts. Configure .env with API keys like ANTHROPIC_API_KEY. Install Chromium via Playwright.

Docker option pulls ghcr.io/gh05tcrew/pentestagent:kali image. Run with environment variables. Local builds use docker compose.

HexStrike installs via scripts/install_hexstrike_deps.sh. Add MCP configs like pentestagent mcp add nmap.

Attack Playbooks

• THP3-style web assessments.
• Reconnaissance to exploitation flows.
• RAG-enhanced domain knowledge.
• Persistent notes and shadow graphs.

Built-in Tools

• Terminal: nmap, sqlmap execution.
• Browser: Playwright automation.
• Notes: loot/notes.json storage.
• Search: Tavily API queries.

TUI shows memory usage. Esc stops agents. /memory checks status.

Integration Benefits

HexStrike MCP exposes pentest tools via mcp_servers.json. CLI lists and tests servers. Stability fixes aid long tasks.

Legal Requirements

• Test only authorized systems.
• Unauthorized access breaks laws.
• Document permission scopes.

FAQ