deserialization-insecure

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly includes and instructs using hard-coded encryption keys (e.g., Shiro default AES keys) and shows embedding them verbatim into cookies/requests, which requires the model to handle and output secret key values directly.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This content is an explicit offensive exploitation playbook that describes how to craft, deliver, and verify deserialization payloads (ysoserial/phpggc/marshalsec/etc.) to achieve RCE, data exfiltration, and remote classloading from attacker-controlled servers.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I found literal, high-entropy AES keys in the SKILL.md content. These base64 strings (e.g., kPH+bIxk5D2deZiIxcaaaA==, wGJlpLanyXlVB1LUUWolBg==, 4AvVhmFLUs0KTA3Kprsdag==, Z3VucwAAAAAAAAAAAAAAAA==) are known default Apache Shiro keys (SHIRO-550/CVE-2016-4437) and qualify as real secrets because they can be used to forge/decrypt rememberMe cookies. They are not placeholders or example markers.

I ignored other apparent values because they are documentation placeholders or low-sensitivity examples: tokens/URLs like TOKEN.collab.net, BURP_COLLAB, UNIQUE.burpcollaborator.net, attacker.com, placeholders like DECRYPTION_KEY_HEX / VALIDATION_KEY_HEX / BASE64_PAYLOAD, env var names (WEBSITE_AUTH_ENCRYPTION_KEY), and command snippets — none are literal high-entropy credentials that grant access.