linkfox-jiimore-get-niche-info-by-keyword

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

• Third-party content exposure detected (medium risk: 0.30). The skill executes a user-supplied keyword via POST /jiimore/getNicheInfoByKeyword and then ingests the returned JSON fields (including strings like nicheTitle/translationZh) into the runtime (stdout and saved files), creating indirect prompt-injection exposure from the API response content that is influenced by the outsider-authored keyword submission.

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

• Secret detected (high risk: 1.00). I scanned the provided files for high-entropy literal values that could be usable credentials. The only candidate that meets the "high-entropy literal" definition is the base64-like constant assigned to _LOGIN_FIXED_UID in scripts/onboarding.py. It is a non-obvious, random-looking string ("eyJhX2lkIjoiNmEyMmM4YjA1YmM5MTZhIiwiZF9pZCI6IiJ9") used as a default UID header when a user_id is not available; such a value can act as an authentication-related header and therefore may be sensitive.

No other hardcoded API keys, private/PEM blocks, or similar high-entropy tokens were found. Environment-variable placeholders (e.g., LINKFOX_AGENT_API_KEY) and example curl uses of $LINKFOXAGENT_API_KEY are documentation placeholders and were ignored per the rules.