The Agent Skills Directory

[SAFE]: The skill instructions and associated reference files are focused entirely on the stated purpose of social media copywriting for affiliate marketing. All operations, including web research, are contextually appropriate.
[PROMPT_INJECTION]: The skill includes self-validation steps to ensure compliance with formatting and FTC disclosure rules. These are benign instructions aimed at maintaining output quality and do not attempt to bypass core AI safety filters.
[INDIRECT_PROMPT_INJECTION]: The skill utilizes web search and browsing capabilities to ingest product data and testimonials from the internet, which represents a potential attack surface for indirect prompt injection. However, this behavior is a core part of the skill's primary research function and is assessed as low risk.
Ingestion points: The web_search and web_browse tools are called in SKILL.md (Step 2 and Step 2.5) to fetch external data.
Boundary markers: None explicitly present to separate untrusted web content from instructions.
Capability inventory: The skill can perform web searches and generate formatted text content.
Sanitization: External content is incorporated into the agent's context for research purposes without specific sanitization beyond standard LLM processing.

viral-post-writer