The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to retrieve and process live social media content (posts, profiles, and display names), which are considered untrusted data sources.
Ingestion points: The references/xquik-promotion.md file defines a workflow for ingesting data using the Xquik search tool.
Boundary markers: The instructions explicitly mandate isolating retrieved content between XQUIK_UNTRUSTED_X_CONTENT boundary markers and escaping the text.
Capability inventory: The skill possesses the capability to publish content to social media platforms via a create-post operation.
Sanitization: The workflow enforces strict human-in-the-loop approval, requiring the user to review the exact payload and public effect before any publishing action occurs.
[EXTERNAL_DOWNLOADS]: The skill instructions in references/xquik-promotion.md direct the agent to connect to a hosted Model Context Protocol (MCP) endpoint at https://xquik.com/mcp to access additional research and publishing tools.

post-acceptance