linkfox-sorftime-walmart-category-market

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

• Third-party content exposure detected (medium risk: 0.30). 该 skill 在运行时通过 POST /sorftime/walmart/categoryMarket 调用外部数据源接口，并把上游返回的 data.value（包含市场/类目/Best Seller等文本字段）直接读入并序列化落盘与摘要输出，因此会让外部服务的响应内容进入上下文。

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

• Secret detected (high risk: 1.00). I scanned the skill files for literal, high-entropy values that could act as reusable credentials.

Flagged:

• scripts/onboarding.py contains a hardcoded base64url string assigned to _LOGIN_FIXED_UID ("eyJhX2lkIjoiNmEyMmM4YjA1YmM5MTZhIiwiZF9pZCI6IiJ9"). The surrounding comment says "登录链路固定 uid header（实测可用）" (fixed uid header, tested usable) and the value is used as a UID header fallback. Because it is a literal token-like value (not a placeholder) and is explicitly described as a usable fixed header, it qualifies as a hardcoded credential.

Ignored (not flagged) examples and reasons:

• Environment variable names and placeholders (e.g., LINKFOX_AGENT_API_KEY, LINKFOXAGENT_API_KEY, API_KEY="${...}") are documentation/configuration references — placeholders, not secrets.
• Example request bodies and nodePath samples (e.g., "4044_623679_...") are structural/sample data, not credentials.
• Other literals (user-agent strings, numeric device IDs, short hex fragments used in examples, default strings instructing how to set env vars) are not high-entropy secrets or are clearly configuration/examples.

Therefore there is one hardcoded, possibly active credential present.