The Agent Skills Directory

External Resource Retrieval: The skill is designed to fetch a JSON catalog and subsequent skill instructions from official repository paths under the Google organization on GitHub. These operations are used to provide up-to-date functionality without preloading a large volume of data.- Command Execution: The skill utilizes standard command-line utilities such as curl, wget, jq, and node to retrieve and process the remote catalog. These tools are employed for their intended purpose of data retrieval and parsing.- Security Best Practices: The instructions contain explicit warnings against disabling security features, specifically prohibiting the use of insecure flags (like curl -k or -SkipCertificateCheck) when fetching remote content. This helps maintain the integrity of the instructions the agent follows.- Indirect Prompt Injection Surface: By design, this skill ingests and follows instructions fetched from external files. While this creates a theoretical surface for indirect prompt injection, the skill mitigates this by restricting its lookups to a specific, verified vendor repository.

finding-google-skills