The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the wmill CLI (a tool provided by the vendor, windmill-labs) to perform actions such as script preview, script run, and generate-metadata. These commands are used to manage the lifecycle of scripts within the workspace.
[DYNAMIC_EXECUTION]: The skill's primary purpose is to enable the agent to author PHP scripts and execute them using wmill script preview. The instructions include a warning that this command executes the script code and may cause side effects, advising the agent to confirm execution intent with the user.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of script parameters and metadata fields (like prompt_for_ai) which could potentially contain malicious instructions.
Ingestion points: Script input parameters passed to the main function and the prompt_for_ai field within script schemas (SKILL.md).
Boundary markers: Not explicitly defined for parameter values, though the skill uses structured PHP function signatures.
Capability inventory: The agent can execute wmill CLI commands and run PHP scripts which may have access to resources or perform network operations.
Sanitization: The skill promotes the use of Windmill resources for sensitive data, which prevents hardcoding secrets, but does not mandate specific sanitization for standard string inputs.

write-script-php