AI Agents
pentagi
Autonomous AI agents for penetration testing — self-hosted, multi-agent, with a full web UI
★ 25.3k starsMIT
View guide →Automated penetration-testing agent framework powered by LLMs, published at USENIX Security 2024.
PentestGPT is an open-source ai agents project by GreyDGL (GreyDGL/PentestGPT on GitHub). It has 15,834 GitHub stars and 2,761 forks, is released under the MIT license, and was last updated on GitHub on 14 July 2026.
Only test systems you own or are explicitly authorized to assess; note the target address and any instructions for the engagement.
Clone the repo and run make install, which uses uv to sync dependencies; a self-contained Docker image is also provided.
Install and authenticate the Claude Code or Codex CLI for autonomous runs, or configure a provider for the interactive multi-LLM mode.
Point pentestgpt at a target in CTF or pentest mode, watch the staged pipeline work, and resume later from a saved session if needed.
The agent works through staged phases and feeds each stage's findings into the next, for both CTF and penetration-test workflows.
Drives Claude Code or Codex to run tools and reason through a target without human intervention at each step.
Covers Web, Crypto, Reversing, Forensics, PWN and Privilege Escalation challenge categories.
Save and resume penetration testing sessions instead of restarting long engagements from scratch.
Tracks steps in real time with live activity updates as the agent works through a challenge.
The modernized legacy mode supports OpenAI, Anthropic, Gemini, DeepSeek, xAI, Qwen, Moonshot and local Ollama providers.
Clone PentestGPT and enter the project directory.
git clone https://github.com/GreyDGL/PentestGPT.git
cd PentestGPTInstall the project with the Makefile target, which runs uv sync (Python 3.12+ and uv required).
make installStart the agent against a target you are authorized to test; CTF mode is the default.
pentestgpt --target 10.10.11.234It is maintained by GreyDGL and contributors, and the underlying research was published at USENIX Security 2024.
The code is MIT-licensed and free. Only use it against systems you own or are authorized to test.
Usually: autonomous mode needs an authenticated Claude Code or Codex CLI, and interactive mode uses your own provider keys, though local Ollama is supported there at no API cost.
PentestGPT adds a staged pipeline, tool execution, live walkthrough tracking and session persistence around the model, instead of one-off chat answers you must act on manually.
Clone the repo, run make install, authenticate a backend CLI, then run pentestgpt --target <ip> against an authorized lab target.
Autonomous AI agents for penetration testing — self-hosted, multi-agent, with a full web UI
AI Agent Assistant & development framework that integrates lots of IM platforms, LLMs, plugins and AI feature, and can be your openclaw alternative. ✨
NVIDIA's security scanner for AI agent skills - check before you install
Retired agent framework: its prompts and guides live on in successor repos.
A framework for agent nexuses where intelligence emerges from interaction.
Agent framework for the JVM. Pronounced Em-BAY-bel /ɛmˈbeɪbəl/
The most-starred other AI Agents projects in the LibreAI catalogue — the realistic shortlist if PentestGPT is not the right fit.
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
Free and open — grab the source, star the repo, and start creating.