Hysen LabsPentest-Swarm-AI
Autonomous penetration testing using a swarm of AI agents. Orchestrates recon, classification, exploitation, and reporting specialists with ReAct reasoning — supports bug bounty, continuous monitoring, and CTF modes. Built with Go, Claude API, and 7+ native security tools.
Pentest Swarm AI runs security tests with a team of agents
Pentest Swarm AI is an open source penetration testing tool written in Go that coordinates several AI agents to perform reconnaissance, classification, exploitation, and reporting. The project is licensed under AGPL 3.0.
A swarm instead of one agent
The tool describes itself as the first open source pentesting tool built on a real swarm, meaning specialized agents work together rather than a single agent running steps in a row. The README contrasts a swarm with a plain multi agent setup, where several models are simply chained. The design aims to mirror how a human security team splits work.
How a campaign runs
The project organizes the work into phases: reconnaissance, classification, exploitation, and reporting. Each specialist agent handles one phase and passes results along. The README states it supports bug bounty, continuous monitoring, and CTF modes, which are authorized contexts for security testing. It is built with Go and can use the Claude API or a local Ollama model.
Status and scope
The README marks the project as alpha, so the feature set is still changing. It bundles seven or more native security tools that the agents drive during a campaign. With more than 2,200 stars, it sits among the more watched AI assisted security projects on GitHub, though users should treat it as experimental.
Editorial conclusion
Pentest Swarm AI is written in Go, licensed under AGPL 3.0, and the README notes it is in an alpha state while supporting both the Claude API and local Ollama models.
Community notes