linkfox-amazon-store-fba

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

• Third-party content exposure detected (medium risk: 0.30). Skill runtime path scripts/_fba_runner.py builds the outgoing request from user-provided JSON (query, queryString, and requestBody) and forwards it via POST ${LINKFOX_TOOL_GATEWAY}/spApi/developerProxy, so outsider-authored free text provided as these fields can be ingested at runtime (into the proxy call body/query).

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

• Secret detected (high risk: 1.00). I scanned the files for high-entropy literal values that could be used to access services.

Flagged:

• scripts/onboarding.py lines 48-49 define _LOGIN_FIXED_UID = "eyJhX2lkIjoiNmEyMmM4YjA1YmM5MTZhIiwiZF9pZCI6IiJ9". This is a base64/JWT-like, high-entropy literal used as the "登录链路固定 uid header（实测可用）" (fixed uid header for login) — i.e., it is used in the auth/login path and therefore can function as an access-related token. I consider this a real secret.

Ignored (reasons):

• SKILL.md export LINKFOXAGENT_API_KEY="" and other placeholders like "A1...", "B0...", "" — documentation placeholders per rules (not real keys).
• Environment variable names (e.g., LINKFOX_AGENT_API_KEY, LINKFOXAGENT_API_KEY) and default URLs — not secrets.
• Other strings in code (UA, TAG, numeric IDs) are low-entropy or non-credential values and are ignored.