attack-mutator

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The mutator explicitly generates variations of the provided "original" input (including base64/ROT13/unicode/zero-width encodings and paraphrases), so if that input contains an API key, token, password, or other secret the model will reproduce and emit those secret values (possibly encoded), enabling exfiltration.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). Suspicious: this is an untrusted GitHub repository (unknown/individual account) that explicitly provides an "Attack Mutator" with obfuscation and bypass techniques—a dual‑use tool likely to facilitate malware/abuse rather than a vetted vendor release.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This content explicitly documents and instructs on evasion, obfuscation, and jailbreak techniques (zero-width injection, homoglyphs, base64/ROT13 wrapping, "best-of-n" jailbreaking) and provides examples for bypassing filters and varying attack vectors, making it intentionally enabling of adversarial and malicious behavior.