The Agent Skills Directory

[SAFE]: The skill uses established platform-specific CLI commands (wmill) to facilitate development workflows. These include script preview for local execution and sync push for deployment, which are standard operations for the Windmill platform.
[COMMAND_EXECUTION]: The instructions guide the agent on how to use wmill script preview and wmill generate-metadata. These commands are used to run local scripts and update configuration files (.lock, .script.yaml) respectively, which is appropriate for a developer-oriented tool.
[DATA_EXFILTRATION]: While the skill describes streaming query results to S3 and downloading S3 objects, these are documented features of the Windmill platform (using the -- s3 directive and s3object type). These operations occur within the user's controlled environment and do not represent unauthorized data exfiltration.
[PROMPT_INJECTION]: The skill provides operational instructions to the agent regarding user interaction (e.g., when to offer a preview). These instructions do not attempt to bypass safety guardrails or override core system behavior.

write-script-postgresql