video-downloader

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). The URLs include an unknown third‑party API domain (redfox.hk) that the skill uses to generate direct download links (and even embeds a public API key in code), which is an untrusted download source that could be used to serve malicious payloads; the remaining links are standard social media/CDN URLs and are not flagged.

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

• Third-party content exposure detected (medium risk: 0.30). In assets/downloader.py, outsider-provided input is read as a user-supplied URL (CLI argument url), validated, then sent to the redfox.hk endpoint (session.post(API_URL, json={"url": url, ...})) where it is used to fetch/parse the target content.

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

• Secret detected (high risk: 1.00). I flagged a hardcoded, high-entropy API key literal found in the code. Specifically, assets/downloader.py contains PUBLIC_API_KEY = "ak_b45b6a6881f4400fb321428947eb6661" — this is a non-trivial, random-looking token that can be used to access the upstream API and therefore meets the definition of a secret.

I ignored other occurrences that look like placeholders or documentation examples:

• Strings like "ark_你的密钥", "ark_xxxxx", and mentions of the environment variable name REDFOX_API_KEY are documentation placeholders or env var names (explicitly allowed to ignore).
• API endpoints (e.g., https://redfox.hk/...) and other configuration text are not secrets.

Because the public key literal is present in source, I mark this as a real secret occurrence.