The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes script metadata and input schemas which can include a prompt_for_ai field. This field is intended to provide instructions to the agent when choosing inputs, representing a surface where untrusted data could influence agent behavior.
Ingestion points: Script input parameters and the prompt_for_ai metadata field in .script.yaml files.
Boundary markers: No specific delimiters or safety instructions are defined for the interpolation of the prompt_for_ai content.
Capability inventory: The agent can execute wmill CLI commands including script preview, script run, and sync push which interact with the local filesystem and the Windmill platform.
Sanitization: The instructions do not specify sanitization or validation for the prompt_for_ai field before it is used by the AI agent.
[COMMAND_EXECUTION]: The skill uses the wmill CLI tool to perform operations. The wmill script preview command executes the script code locally to validate functionality, which may result in side effects as noted in the instructions.
[DYNAMIC_EXECUTION]: The skill is specifically designed to facilitate the creation and runtime execution of C# scripts. It uses the #r "nuget: ..." directive for dynamic package resolution and executes scripts via the wmill CLI during the development and testing workflow.

write-script-csharp