tencentmap-jsapi-gl-skill

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). 该技能要求检测并“静默记录”用户提供的正式 Key 或从网页提取体验 Key，并在示例代码（如 script src 的 key 参数）或后续请求中使用/填入该 Key，迫使模型处理并可能输出秘密值。

---

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The content contains an explicit instruction to "silently record" users' formal API keys (TMAP_JSAPI_KEY) and to check environment or user-provided keys without prompting, which indicates intentional credential collection/privacy-invasive behavior (possible credential theft / exfiltration), while the rest of the docs are normal API documentation and JSONP examples.

---

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly requires fetching and loading remote JavaScript at runtime (e.g. the Tencent Maps SDK script URL "https://map.qq.com/api/gljs?v=3&key={TMAP_JSAPI_KEY}" and it instructs a WebFetch to "https://lbs.qq.com/webApi/uriV1/uriGuide/uriMobileMarker" to obtain an experience Key), which will execute remote code and is treated as a required runtime dependency.

---

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

• Secret detected (high risk: 1.00). I scanned the document for literal high-entropy credentials and ignored placeholders (e.g., {TMAP_JSAPI_KEY}) and obvious simple example passwords. I found a literal API key-like string used in script src attributes:
• OB4BZ-D4W3U-B7VVO-4PJWW-6TKDJ-WPB77

This is not a placeholder (it is a concrete key value appearing multiple times in example script URLs) and matches the format of a real Tencent Maps API key (structured, non-trivial entropy). No PEM/RSA blocks, JWTs, or other high-entropy secrets were found beyond this API key. Other references to keys are placeholders or parameter names (e.g., servicesk, key={TMAP_JSAPI_KEY}) which I ignored per the rules.

---

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)