amz-competitor-analysis

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow explicitly asks the operating user to “paste” competitor inputs and then “mine competitor reviews” (1–3 star review text); those competitor reviews are outsider-authored free text that the agent would ingest into the LLM context.

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

• Secret detected (high risk: 1.00). I found a WhatsApp group invite URL in the skill: https://chat.whatsapp.com/ILX65p1yWcaIG3c9WGHpTY. The path contains a 24-character token-like string (ILX65p1yWcaIG3c9WGHpTY) that is high-entropy and grants access to the group — i.e., it functions as an access credential. No other API keys, private keys, or PEM blocks were present. The <potential_matches> section was empty, so nothing else to review.

Note: if this invite link is intentionally public (a promotional join link), it may be meant to be shared; but by the scanner definition, it is a literal, usable access token and thus flagged as a secret.