AgentCrash is an open-source security test harness for AI agents. It provides synthetic workplace tools, visual evidence, and regression tests to evaluate prompt injection and related agent failures.
AgentCrash sits in PulseGate's Agent evaluation & testing category. It focuses on testing AI agents for prompt-injection vulnerabilities and unsafe tool behavior before deployment. It is built as an open-source project for AI security engineers and agent developers. AgentCrash is open source under the Apache-2.0 license. AgentCrash is available on the command line, and it can be self-hosted.
Behind AgentCrash is Prathamesh Ops Sudo, and it first shipped in 2026. Development happens publicly on GitHub with 17 commits in the last 90 days. Among its 7 catalogued features are synthetic workplace, security testing, and synthetic tools.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do