The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface typical of content generation tools.
Ingestion points: User-provided <topic> and --languages arguments.
Boundary markers: Not explicitly defined within this orchestrator skill.
Capability inventory: File system write access (creating directories and saving markdown/SEO files) and invocation of sub-skills.
Sanitization: No explicit sanitization is performed on input arguments before they are passed to the generation pipeline.
[COMMAND_EXECUTION]: The skill performs file system management and file creation.
Evidence: The workflow in SKILL.md describes creating a multilingual/ directory and writing various artifacts (Markdown, HTML, XML, JSON).
Security Controls: The skill includes a specific rule: "Output MUST stay inside the project root. Never write outside the cwd.", which limits the scope of potential file system impact.

blog-multilingual