The Agent Skills Directory

[SAFE]: The skill is a comprehensive educational resource for the Cortex Code CLI tool. It contains no executable code, scripts, or external dependencies.
[COMMAND_EXECUTION]: While the tutorial explains and guides the user to use the !command (shell) syntax, this is done strictly for educational purposes to demonstrate the tool's built-in functionality. No malicious or unauthorized commands are automatically executed by the skill itself.
[DATA_EXFILTRATION]: The tutorial references Snowflake table metadata and local file access using the # and @ syntax. These are documented as features of the Cortex Code CLI for user education, with no evidence of sensitive data being harvested or transmitted to external domains.
[PROMPT_INJECTION]: The skill utilizes a structured persona and instructional framework (Explain-Before-Execute) to guide the agent's behavior. No attempts to bypass safety filters or override system instructions were found.
[DYNAMIC_CONTEXT_INJECTION]: The skill mentions and documents the usage of dynamic command execution placeholders (!command). However, it does not use them for malicious purposes such as silent exfiltration or credential harvesting; it serves as a teaching aid for the user.

cortex-code-tutorial