The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands including 'date' (to generate front matter timestamps), 'hugo' (to verify site builds and render draft content), and 'python3' (to run a bundled link-checking script). These are used to manage the project lifecycle on the local filesystem.- [DYNAMIC_EXECUTION]: The 'scripts/blog_links.py' script uses 'subprocess.run' to invoke the 'hugo' CLI tool for site analysis. This is used to query the state of the Hugo site to ensure internal links are valid and target published pages.- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input (topic details, audience level) and interpolates it into writing frameworks stored in 'references/'.* Ingestion points: Untrusted data enters via user requests and repository files. * Boundary markers: The workflow uses template placeholders and explicit framework prompts to constrain the agent's output. * Capability inventory: Includes subprocess calls for 'hugo' and 'date', and file system writes to the blog repository. * Sanitization: Employs a mandatory review loop (Step 9) using the 'write-review' skill and an 'ai-sanitize' step to remove AI-specific patterns and ensure adherence to site-specific rubrics.

write-article