The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill continuously monitors user conversation to detect and register UI/UX concepts. This behavior introduces a surface for indirect prompt injection where malicious instructions could be embedded in the discussed topics. However, the skill includes specific guardrails to mitigate this risk.
Ingestion points: User conversation messages and manual @ux commands analyzed in SKILL.md.
Boundary markers: The 'Guardrails' section explicitly instructs the agent to treat conversation content as data only and ignore any embedded directives or skill invocations.
Capability inventory: The skill utilizes Notion MCP tools for searching, reading, and writing to a database (API-post-search, API-post-page, API-query-data-source, API-patch-page).
Sanitization: Input is normalized (lowercased and special characters removed) before being used as a search key, which provides a basic level of sanitization.

ux-sentinel