ai-teaching-media

No evidence of malicious behavior or sabotage is present. The code is a legitimate animation-rendering template, and the current null DATA placeholder intentionally prevents direct rendering. If generated data can be influenced by an attacker, numerous unescaped fields inserted via innerHTML can cause HTML/script or event-handler injection, and ch.image can enable attribute/resource injection. The GSAP CDN dependency should preferably use Subresource Integrity or a trusted local copy.

SUSPICIOUS. The skill’s capabilities broadly match its teaching-media purpose, but it routes generation through multiple third-party gateway providers and relies on unofficial or unpinned tooling (`edge-tts`, `npx` HyperFrames). This is not fundamentally incompatible with the stated purpose, yet the install trust and credential/data-flow model are broader than a tightly scoped vendor-direct integration.

The script is primarily a legitimate media-rendering pipeline and shows no direct malware indicators. However, it has a significant injection weakness because SILENT_SEG is inserted into Python source without validation, and it executes an unpinned npm package through npx --yes. Use a numeric validation strategy for SILENT_SEG, pin or locally install the renderer, and restrict executable overrides in untrusted environments.

The code is a legitimate media-rendering pipeline and shows no direct evidence of malware. The main security concern is execution of an unpinned npm package via `npx --yes`, which creates a meaningful supply-chain exposure. Unquoted environment-controlled ffmpeg variables and unrestricted output paths are additional hardening concerns. Pin the HyperFrames version and registry/integrity, quote or validate executable variables, and constrain output paths.