The Agent Skills Directory

[SAFE]: The skill instructions define a legitimate workflow for writing blog posts, managing local files, and generating images using platform tools. No evidence of data exfiltration, credential theft, or obfuscation was found.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing content from a researcher sub-agent.
Ingestion points: The agent reads research findings from research/[slug].md.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious instructions embedded within the research data.
Capability inventory: The skill can write files to the local blogs/ directory and invoke the generate_cover tool.
Sanitization: The skill does not implement sanitization or validation of the research content before incorporating it into the final blog post.

blog-post