jwt-cracker

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to accept, echo, embed, or place JWTs and HMAC secrets (e.g., "YOUR.JWT.HERE", "FORGED_TOKEN", "found_secret", "attackersecret") verbatim into commands, code, and Authorization headers, which requires the LLM to handle secret values directly and therefore poses an exfiltration risk.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). Several URLs point to attacker-controlled domains and JKU/JWK injection patterns (attacker.com, evil.com, target.com@attacker.com, redirect to attacker.com) which are commonly used to host malicious JWKS/keys or otherwise deliver/trigger unsafe payloads, so they are suspicious download/host vectors.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This skill explicitly provides step-by-step offensive techniques to forge and abuse JWTs (alg:none, alg confusion, brute-force secrets, kid SQLi/path traversal, jku/jwk injection), host attacker-controlled keys, and lists password wordlists — clearly enabling credential theft and unauthorized access.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow ingests outsider-authored free text via the user prompt (e.g., a provided “string that looks like a JWT” and accompanying test instructions), and the agent then decodes/uses that text as the JWT attack input.