space-markdown-poster
Pass
Audited by Gen Agent Trust Hub on Jul 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts (
scripts/md2img.pyandscripts/md_to_image.py) to process Markdown files and generate images using the Playwright browser automation framework. - [EXTERNAL_DOWNLOADS]: The scripts fetch assets from well-known services, including styles from Google Fonts (
fonts.googleapis.com) and YouTube video thumbnails fromimg.youtube.com,i.ytimg.com, and thenoembed.comservice. - [DATA_EXFILTRATION]: The script
scripts/md2img.pycontains a hardcoded default output directory path (/Users/ugreen/Documents/obsidian/attachments), which exposes the author's local system username (ugreen) to the execution environment. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the rendering of untrusted Markdown content.
- Ingestion points: Arbitrary Markdown content is read from files provided as command-line arguments.
- Boundary markers: There are no delimiters to isolate user content from the rendering logic.
- Capability inventory: The skill uses a headless browser (Playwright) which can execute embedded JavaScript and potentially access local resources.
- Sanitization: The conversion logic only transforms specific Markdown syntax and fails to escape general HTML tags, allowing malicious scripts in the input Markdown to execute during the image generation process.
Audit Metadata