The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents standard infrastructure-as-code workflows using the Pulumi CLI, including project scaffolding (pulumi new), state synchronization (pulumi refresh), and deployment operations (pulumi up, pulumi destroy). It also includes guidance on using pulumi env run to execute commands within a secure environment context.
[EXTERNAL_DOWNLOADS]: Provides instructions for incorporating reusable multi-language infrastructure components from external Git repositories via the pulumi package add command, which is a native feature of the Pulumi ecosystem.
[SAFE]: The skill explicitly discourages the use of static credentials and hardcoded secrets, instead promoting the use of Pulumi ESC for encrypted configuration and OIDC for secure, short-lived cloud provider access.
[PROMPT_INJECTION]: Analyzed for indirect prompt injection risks associated with processing external configuration and dynamic data.
Ingestion points: External data fetched during resource creation (e.g., via fetchExternalData) and configuration values defined in Pulumi ESC YAML files.
Boundary markers: Not explicitly implemented in the provided examples.
Capability inventory: Shell command execution through pulumi up, npm test, and pulumi env run across multiple project files.
Sanitization: Relies on Pulumi's internal serialization and the user's manual validation of the pulumi preview output before deployment.

pulumi-typescript