The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct the user to install the shopino-mcp package using the uvx tool (e.g., uvx shopino-mcp). This package acts as the MCP server that provides the skill's core functionality.
[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes unstructured text from external fashion blog posts and product details, creating a surface for potential indirect prompt injection attacks.
Ingestion points: The sh_blog_post tool fetches blog content as plain text, and search tools like sh_search, sh_browse, and sh_product retrieve external product descriptions.
Boundary markers: There are no boundary markers or specific instructions provided in the skill to delimit external data or warn the agent to ignore embedded instructions within that data.
Capability inventory: The skill is described as read-only; tools are limited to data retrieval and do not provide capabilities for file system writes, shell command execution, or other privileged operations.
Sanitization: No sanitization or filtering of the content retrieved from the external Shopino APIs is specified in the skill's instructions.

shopino