The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted data from WordPress post content.
Ingestion points: Raw content is ingested from SKILL.md through wordpress_extract_builder_content and wordpress_read_page.
Boundary markers: Absent. The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the shortcodes being parsed.
Capability inventory: The agent has write access to the WordPress database via wordpress_update_page and wordpress_update_post.
Sanitization: No explicit sanitization of input content is described, though the skill mandates creating draft duplicates and manual user verification as a security control.
[DATA_EXFILTRATION]: The skill transmits operational metadata and site context to the author's infrastructure.
Evidence: Found telemetry endpoint https://www.respira.press/api/skills/track-usage in SKILL.md.
Context: The telemetry is sent to the official domain of the vendor (respira-press). It includes non-sensitive counts of migrated elements, duration, and site/version context for usage tracking.

migrate-wpbakery-to-gutenberg