The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection by instructing the agent to inspect external media from URLs in references/cases.md using browser or media inspection tools.
Ingestion points: External URLs for goodcase.ai and x.com listed in references/cases.md.
Boundary markers: The instructions include a verification checklist requiring the agent to distinguish between observed evidence and its own recommendations, and to label results as 'prompt-grounded' if media cannot be inspected.
Capability inventory: Usage of browser, image, or video tools to access external content.
Sanitization: The workflow requires a reference contract to explicitly define traits to preserve, replace, or avoid, which acts as a manual filtering step.
[EXTERNAL_DOWNLOADS]: The skill references multiple external URLs for visual examples and creator attribution.
Evidence: The file references/cases.md contains links to goodcase.ai and x.com. These are legitimate references to the creative works of the artist mentioned in the skill description and are necessary for the skill's intended function of style replication.

pov-vlog-presence-by-1jqwyww