The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external media URLs which could potentially serve as a vector for indirect prompt injection if the processed content influences agent behavior or if the server-side processing is exploited.
Ingestion points: mediaUrls parameter in the create_post tool (SKILL.md).
Boundary markers: No explicit boundary markers or "ignore instructions" warnings are defined for the data fetched from external media URLs.
Capability inventory: The skill has the capability to create, update, and delete posts on X, and perform network operations to the Publora API across all its tools (SKILL.md).
Sanitization: No specific sanitization or validation of the external media content is described within the skill instructions.
[SAFE]: All external domains and services mentioned, such as publora.com, mcp.publora.com, and api.publora.com, are owned and operated by the skill's author, Publora, representing normal vendor functionality.
[SAFE]: The use of placeholders like sk_YOUR_API_KEY for authentication is a standard and safe practice for providing setup instructions.

x-post