The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform broad filesystem scans using shell commands such as find ~ and dir /s to dynamically locate sub-skills and configuration files. This discovery mechanism involves invasive filesystem operations across the user's home directory.
[PROMPT_INJECTION]: The skill includes a binary-encoded ZIP archive (hand-drawn-infographic-updated.skill) that contains supplementary instructions and documentation (SKILL.md, design-guide.md). Embedding instructions in binary format obscures them from plain-text security audits and automated scanners.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests large amounts of untrusted user-provided content (e.g., articles for rewriting) to drive its internal analysis logic and image prompt generation without using boundary markers or sanitization.
Ingestion points: User-provided text is saved to source.md and analyzed in analysis.md as part of the content creation workflow.
Boundary markers: Absent. No clear instructions are provided to the agent to treat the input as data only or to disregard any embedded instructions.
Capability inventory: The skill has the ability to execute shell commands, manage local files, and call interactive user-facing tools.
Sanitization: Absent.
[EXTERNAL_DOWNLOADS]: The skill fetches styling assets and font resources from well-known public CDN services including jsDelivr, Cloudflare, and Google Fonts to generate HTML previews. These resources originate from established and reputable providers.

xiaohongshu-creation-workflow