The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions frequently require the agent to use WebSearch and WebFetch to ingest documentation from docs.medusajs.com. This external data ingestion combined with the agent's ability to write code and execute shell commands creates a surface for indirect prompt injection.
Ingestion points: Multiple instructions in SKILL.md to fetch live documentation and search for testing patterns on docs.medusajs.com.
Boundary markers: The skill does not provide clear delimiters or instructions to the agent to ignore potentially malicious directions embedded in the fetched documentation.
Capability inventory: The skill is granted Write, Edit, and Bash tools, which could be leveraged if the agent follows malicious instructions from a compromised external source.
Sanitization: There is no evidence of sanitization or validation mechanisms for the content fetched from the web.
[EXTERNAL_DOWNLOADS]: The skill makes several requests to fetch documentation from docs.medusajs.com and mentions standard testing dependencies.
Evidence: References to fetching resources from https://docs.medusajs.com/resources/medusa-testing and searching the official docs for integration test utilities and setup.

medusa-testing