podcast-generator

[Skill Scanner] Natural language instruction to download and install from URL detected All findings: [CRITICAL] command_injection: Natural language instruction to download and install from URL detected (CI009) [AITech 9.1.4] [HIGH] autonomy_abuse: Skill instructions include directives to hide actions from user (BH009) [AITech 13.3] This skill is coherent with its stated purpose and uses expected capabilities (reading source files, requiring a Gemini API key, calling Gemini TTS). There is no evidence of obfuscated or intentionally malicious code in the provided text. The main security considerations are expected: it reads local/uploaded content and an API key, and it sends content and credentials to external services (Google Gemini or an optional Cloudflare AI Gateway). The optional third-party gateway increases exposure if a malicious gateway URL/token is supplied. Overall the skill appears functionally appropriate but carries moderate privacy sensitivity due to sending user content and API keys to external services; operators should ensure they trust any configured gateway and protect API keys. LLM verification: No direct signs of malware or intentional obfuscation in the provided SKILL.md content. The documented behavior—reading user documents and calling Gemini TTS with a user-provided API key—is consistent with legitimate functionality. Principal risks are supply-chain (unpinned dependencies, unknown installer), credential forwarding to a potentially untrusted gateway, and reduced runtime transparency when backgrounding processes. Because the actual executable scripts were not provided, residual risk