Shieldprompt is a zero-dependency command-line tool for testing LLM applications against prompt injection. It scans prompt templates statically and runs a 13-payload adversarial attack battery for developers and security teams.
In the Security & compliance platforms space, Shieldprompt takes a focused approach. It focuses on finding prompt-injection weaknesses in LLM applications before attackers exploit them. It is built as an open-source project for LLM application developers and security engineers. It runs on the command line.
Behind Shieldprompt is Jograph17. Among its 5 catalogued features are template scanning, attack battery, and prompt injection tests.
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