The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it is designed to ingest and analyze untrusted external content as part of its primary marketing functions.
Ingestion points: In the 'Tool Integrations' section of SKILL.md, the instructions guide the agent to fetch creator posts and analyze content style/engagement using tools like social-fetch and watch-video.
Boundary markers: The instructions do not explicitly mandate delimiters or instruction-bypass warnings for the content retrieved from social media platforms or influencer media kits.
Capability inventory: The skill's logic is focused on content analysis and strategy drafting. It does not contain instructions for high-risk system operations, though the agent's overall environment may provide them.
Sanitization: There is no specific requirement for the agent to sanitize or filter the fetched external data before processing it for marketing insights.
Contextual Assessment: This vulnerability surface is inherent to the intended use case of vetting external talent and is not indicative of malicious intent. The risk is considered standard for agents interacting with web content.

influencer-marketing