video-frame-extractor

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt embeds a literal API key and instructs setting API keys as environment variables, which exposes a secret-like value that could be copied verbatim into outputs or commands.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The code embeds and/or defaults to a hardcoded third‑party API key and automatically encodes/uploads user images to external endpoints (api.coze.cn and configurable vision APIs), enabling unintended data exfiltration and use of the maintainer's account—an intentional backdoor/credential abuse pattern.

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

• Secret detected (high risk: 1.00). I checked the documentation and scripts for literal, high-entropy credentials. The string "cztei_qHZQ0A5OSJjsmfZWmVb8bqu2BTbtB240YGbDYLhZpsIr8jER4aL4Aevyii8rnKfNs" is a long, random-looking API key (high entropy) and is hardcoded in both the SKILL.md example and in scripts/coze_bot_client.py as a default API_KEY. That meets the definition of a secret.

I ignored placeholders like "your_api_key" and other example values (these are explicit documentation placeholders and not real secrets). The numeric Bot ID is an identifier (present as a default) but is not high-entropy; I did not treat it as the primary secret — the API key is the actual credential to flag.