linkfox-seerfar-ozon-keyword-mining

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

• Third-party content exposure detected (medium risk: 0.30). runtime path ingests outsider-authored free text via the keyword field supplied by the user into POST /seerfar/ozon/keywordMining (called by scripts/seerfar_ozon_keyword_mining.py:call_api()), which then returns mined keyword strings/fields the LLM may read for summarization/output.

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

• Secret detected (high risk: 1.00). I looked for literal, high-entropy values that could function as credentials. The only candidate that meets that definition is the base64url string assigned to _LOGIN_FIXED_UID in scripts/onboarding.py. This value is not a placeholder (it is a concrete encoded token) and the code comment says it is a "fixed uid header (实测可用)". The value is injected into request headers (uid) when no access_token is present, so it can be used in requests and therefore represents an operational credential-like secret.

No other high-entropy API keys, JWTs, or private keys are present. Environment variable names, example IDs in responses, short/simple passwords, and static non-secret strings (user agents, URLs, numeric device ids) are intentionally ignored per the rules.