design-studio

The security audit of the 'design-studio' skill concludes that the skill is safe for use. The analysis focused on the following areas:

• [EXTERNAL_DOWNLOADS]: The skill fetches legitimate assets (images) from Wikimedia Commons via a dedicated Python script (fetch_images.py). It also references well-known and trusted CDNs (unpkg.com, googleapis.com, jsdelivr.net) for loading frontend libraries like React, Babel, and Lucide icons. These operations are essential for the skill's functionality and target reputable domains.
• [COMMAND_EXECUTION]: Several Node.js and shell scripts (e.g., narrate-pipeline.mjs, add-music.sh) execute local instances of ffmpeg and ffprobe to perform audio and video processing. These commands are constructed using parameters derived from the local environment and intended inputs, presenting no evidence of arbitrary or malicious execution.
• [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes user-provided markdown files for narration and reverse-engineers designs from user-supplied references. However, the risk is assessed as low because the instructions are tailored to treat these inputs as data for processing (e.g., text for Text-to-Speech) rather than instructions for the agent. The skill's primary function is to transform this data into specific outputs.
• [DYNAMIC_EXECUTION]: A script (export_deck_pptx.mjs) dynamically loads a local module (html2pptx.js) using a computed path. This is a standard modularization pattern within the skill's own directory structure and does not involve executing external or untrusted code.
• [CREDENTIALS_SAFE]: The skill uses environment variables (.env) for managing API keys for external services (e.g., Doubao TTS), which is a recommended security practice. There are no hardcoded secrets found in the provided files.