The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references a URL on the tool's official website (glasser.ai) for installation instructions and skill updates. This is treated as a safe reference to a well-known service domain.
[COMMAND_EXECUTION]: The skill utilizes a local CLI tool (glasser) to search for endpoints, inspect schemas, and execute data requests. All commands are standard for the tool's purpose and do not include dangerous or obfuscated patterns.
[CREDENTIALS_UNSAFE]: The skill explicitly instructs users and agents to use environment secret managers for GLASSER_API_KEY and warns against writing keys to files, logs, or command arguments, which aligns with security best practices.
[DATA_EXFILTRATION]: While the tool sends data to third-party providers, it includes explicit instructions to inform the user which provider is receiving the request and to minimize the data sent. No unauthorized exfiltration patterns were detected.

glasser