Introducing Burp AT: agentic AI, built on two decades of Burp Suite
PortSwigger has released Burp AT, an agentic AI tool integrated into Burp Suite Professional that automates pentesting tasks while maintaining human control. The architecture enforces security boundaries: agents operate through Burp’s battle-tested tools and can access project context (traffic, scope, issues), but execute only actions that Burp’s scope and approval rules permit—agents propose, Burp enforces. Skills are purpose-built from PortSwigger Research methodology rather than general LLM knowledge. In closed beta, a pentester used it to analyze 66,000 lines of minified JavaScript in four days and surfaced a critical vulnerability that would have gone untested for at least a year. The approach prioritizes auditability and reproducibility: all requests and evidence flow through Burp’s tooling so findings can be independently verified.