ocr-to-en-word

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly instructs users to "直接告诉AI你的密钥" (APP_ID, API_SECRET, API_KEY) and states the AI will automatically save them verbatim into a .key file and prompt for replacement keys on failures, which requires the model to receive and output secret values directly.

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

• Secret detected (high risk: 1.00). I scanned the provided skill content for literal, high-entropy credentials. The .key file contains concrete values (not placeholders):

{ "APP_ID": "ce9b7a0e", "API_SECRET": "YzY3MzNlMmQxZjMxNWMyNjBiZGEyZTkw", "API_KEY": "b2fd97aabe4a132084e5d018f362354c" }

• APP_ID is a short identifier but paired with API_SECRET and API_KEY that are random-looking (base64-like and 32-hex-like respectively). These match the definition of secrets (API credentials) and are directly usable for authentication in the scripts.
• Other values in the docs are placeholders (e.g., "your_app_id", "your_api_key") and example strings; those are ignored per rules.

Therefore this is an actual secret leak.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)