The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill requires the agent to inspect external media and web pages, which could host malicious instructions intended to influence the agent's output or behavior.
Ingestion points: references/cases.md directs the agent to access and analyze content from goodcase.ai and x.com using browser or media inspection tools.
Boundary markers: The instructions lack explicit safety delimiters or warnings to ignore instructions that might be embedded within the external metadata or page content.
Capability inventory: The agent is tasked with browsing, media analysis, and generating complex text-based prompt artifacts based on the fetched data.
Sanitization: There is no requirement or logic provided for sanitizing or filtering content retrieved from these external sources.
[EXTERNAL_DOWNLOADS]: The skill depends on resources from non-whitelisted external domains for its core functionality.
Evidence: The file references/cases.md contains several links to goodcase.ai and media.goodcase.ai. The workflow in SKILL.md explicitly instructs the agent to "Inspect the anchor's finished media" via these external links.

pov-vlog-presence-by-v2wqfs