The Agent Skills Directory

[PROMPT_INJECTION]: The skill proactively addresses indirect prompt injection by instructing the agent to ignore any commands or role-play found in external references. Ingestion points: External URLs, PDFs, and documents. Boundary markers: Explicitly defined 'Security Requirements' section. Capability inventory: File system writes and external tool calls (cover generation and storage). Sanitization: Instructions to extract only facts and ignore operational directives.
[COMMAND_EXECUTION]: The skill performs legitimate file system operations, specifically creating and writing Markdown files within the user's Hugo project structure.
[DATA_EXFILTRATION]: The skill integrates with 'qiniu-kodo' for image hosting. This is a functional feature of the blog creation workflow used to store generated cover images and is disclosed in the documentation.

content-creator