pulumi-cli

name	pulumi-cli
description	Execute Pulumi CLI commands for stack management, infrastructure deployments, and API interactions. You MUST load this skill when working with the pulumi command.
license	MIT

Pulumi CLI

When to Use

• Executing core Pulumi commands like pulumi up, pulumi preview, or pulumi destroy.
• Managing Pulumi stacks with pulumi stack and configuration with pulumi config.
• Interacting directly with the Pulumi Cloud REST API using pulumi api.

When Not to Use

• Writing or refactoring Pulumi infrastructure code (use language-specific skills like python or typescript).
• Fetching documentation or schemas (use pulumi-docs instead).

Core Principles

• Non-Interactive Execution: Autonomous agents MUST always use the --non-interactive flag to prevent the CLI from hanging while waiting for human input.
• JSON Output: Use --output=json where supported (especially with pulumi api) for stable and parsable outputs that agents can easily consume.
• Stable Exit Codes: Rely on Pulumi CLI's exit codes for stable error handling in automation scripts.

Step-by-Step Workflows

Deploying Infrastructure

1. Preview Changes: Always run a preview before deploying to understand the impact. pulumi preview --non-interactive
2. Deploy Changes: Deploy the infrastructure changes automatically. pulumi up --non-interactive --yes
3. Destroy Infrastructure: Tear down the stack entirely when instructed. pulumi destroy --non-interactive --yes

Managing Stacks

1. List Stacks: pulumi stack ls
2. Select Stack: pulumi stack select <stack-name>
3. Show Fully Qualified Stack Names: pulumi stack ls -Q

Interacting with Pulumi API (Agent Workflow)

1. List API Endpoints: pulumi api list --output=json
2. Call API Endpoint: pulumi api /api/user

Common Pitfalls

• Interactive Hangs: Running pulumi up or pulumi destroy without --non-interactive and --yes will cause the agent session to hang waiting for user confirmation.

What to Avoid

• Avoid modifying Pulumi state directly unless absolutely necessary. Rely on standard deployments where possible.
• Do not use interactive commands in agent scripts. Always supply all required flags and use --non-interactive.

Related Skills

• pulumi-docs: You MUST load this skill when retrieving Pulumi documentation or schemas.