cookie-analysis

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs inserting session cookie values and secret keys verbatim into commands (e.g., curl -b 'session=...', flask-unsign --cookie ''), so an agent would need to handle and output sensitive secrets directly.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This content is an explicit offensive methodology for forging session cookies and bypassing web authentication (unsigned base64 cookies, Flask signed cookies, JWT weaknesses, padding-oracle attacks and brute-forcing secrets), enabling unauthorized access and privilege escalation and therefore constitutes deliberate malicious guidance.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs fetching cookies from an external target (Step 1: "curl -sI http://TARGET/ | grep -i set-cookie") and then decodes and interprets those untrusted cookie values to choose follow-up attack actions, so third-party content can directly influence tool use.