guided-testimonial-video

[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] supply_chain: Installation of third-party script detected (SC006) [AITech 9.1.4] No direct malicious behavior or hard-coded backdoors are present in the provided skill text. The workflow requires actions that are normal for its purpose but introduce supply-chain and data-exposure risks: (1) installing and running npm packages (npx helios), (2) fetching arbitrary URLs for brand research, and (3) using a user-provided ELEVEN_LABS_API_KEY to call an external API. Recommend explicit user consent before any package installation or API key usage, pin and verify package versions and sources, restrict/validate fetched URLs, avoid reading repository secrets, and ensure the environment securely handles API keys (no logging or persistent storage). Treat the skill as functional but requiring caution and operational controls. LLM verification: The fragment describes a coherent high-level workflow for producing brand-aligned testimonial videos using external services (ElevenLabs, Helios) and local rendering. However, it presents elevated supply-chain and runtime risks due to unpinned dependencies, external downloads, and insecure handling of API keys. Mitigations include pinning dependency versions, validating and restricting external tool sources, implementing secret management with rotation and access controls, and clearly documentin