The Agent Skills Directory

[DATA_EXPOSURE]: The skill provides examples of authenticating with the Medusa API using JWT tokens and API keys. All examples use safe placeholders such as sk_test_xxxx..., pk_xxxx..., and REDACTED. It correctly instructs users to handle secrets in their own environment rather than hardcoding them.
[COMMAND_EXECUTION]: The documentation includes standard shell commands (curl, npx) as examples for developers to interact with the Medusa platform. These are clearly presented as instructional recipes and installation commands for related tools from the same vendor (sales-skills).
[EXTERNAL_DOWNLOADS]: The skill references official Medusa documentation and repositories on GitHub (github.com/medusajs). It also mentions community plugins for expanding functionality. These references are to well-known services and trusted organization repositories, handled neutrally as part of the platform's ecosystem.
[INDIRECT_PROMPT_INJECTION]: As a technical helper, the skill is designed to ingest and process data from the Medusa API. While this creates a theoretical surface for indirect prompt injection (e.g., if an API response contained malicious instructions), the skill follows developer best practices by focusing on specific resource IDs and structured JSON responses rather than raw natural language interpolation.

sales-medusa