casefold-fuzz is a lightweight Python library that generates multiple evasion variants from a single prompt injection string. It produces casefold, confusable, invisible-character, leetspeak, and spacing mutations to help developers stress-test their LLM guardrails and security filters. Distributed as a pure-stdlib package under Apache-2.0, it is ideal for security researchers and AI application developers building robust defenses against prompt attacks.
In the LLM eval & observability space, casefold-fuzz takes a focused approach. It focuses on testing LLM guardrails against prompt injection, homoglyph, and evasion attacks by generating mutation variants. casefold-fuzz is an open-source project aimed at developers. casefold-fuzz is open source under the Apache-2.0 license. casefold-fuzz is available on the command line.
Fevzi Ege Yurtsevenler builds and maintains casefold-fuzz, and it first shipped in 2026. Development happens publicly on GitHub with 3 commits in the last 90 days. Key capabilities include Prompt Injection Fuzzing, Homoglyph Generation, and Leetspeak Variants.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Same category — not a similarity match