The Agent Skills Directory

[COMMAND_EXECUTION]: The instructions include a step to verify image availability using curl -sI, which involves executing a shell command with external URLs.
[EXTERNAL_DOWNLOADS]: The skill fetches assets and data from well-known services such as Pixabay, Unsplash, Pexels, and YouTube.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection through its data ingestion processes.
Ingestion points: Web search results, external statistics, and user-provided NotebookLM documents.
Boundary markers: No explicit delimiters are used to separate researched content from agent instructions.
Capability inventory: The agent can execute curl commands and write file content.
Sanitization: No data sanitization or validation is specified for the external content before its use in the final article.
[SAFE]: The skill exhibits no signs of obfuscation, persistence mechanisms, or credential theft. The tools used are consistent with the skill's stated purpose of content research and generation.

blog-write