pulumi · Pulumi
Pulumi CLI Agent 使用指南
通过结构化输出和明确 Stack 控制预览并应用基础设施程序。
官方工具操作风险: R0–R3docs-verified
为 Agent 安装
选择与运行环境匹配的官方安装方式。在团队或 CI 环境中固定版本，并先运行版本检查。
用于稳定自动化的结构化输出
优先使用机器可读格式，并把 stdout 作为结果、stderr 作为诊断信息分别处理。
R0–R3 命令风险指南
风险按单条命令判断。R0 是本地或远程只读，R1 是可逆的本地写入，R2 会改变远程状态，R3 可能造成不可逆或生产级影响。
只读不等于可公开
R0 只表示命令不更改本地或远程状态。只读命令仍可能返回令牌、身份信息、配置或生产数据；只展示完成任务所需的最少内容，不得写入日志、Prompt 或提交内容。
Agent Readiness 评分依据
适配度描述 Agent 操作工具的稳定程度，不代表所有命令都安全，也不替代独立执行测试。
文档证据对应的 Agent Readiness 为 87/100；尚未记录本地执行测试。
生成 Skill 或 Agent 策略
选择目标 Agent 和安全模式，生成包含安装、允许命令、确认边界与证据说明的可复制产物。
生成结果预览
SKILL.md
---
name: pulumi-agent-workflow
description: Use Pulumi CLI for infrastructure as code, stack previews, typed cloud programs with explicit command risk and evidence boundaries.
---
# Pulumi CLI agent workflow
Use this skill when the task needs infrastructure as code, stack previews, typed cloud programs, deployment history.
## Evidence boundary
- Registry confidence: `docs-verified`
- Documentation checked: `2026-07-10`
- Locally tested version: `not tested`
- Do not describe this CLI as locally verified until its commands have actually been executed in an isolated environment.
## Executed smoke checks
- No local execution record is available.
## Installation
- Homebrew (macos, linux): `brew install pulumi/tap/pulumi`
- winget (windows): `winget install Pulumi.Pulumi`
## Authentication
- Methods: Pulumi access token, DIY backend, provider credentials
- Secret environment variables: `PULUMI_ACCESS_TOKEN`, `PULUMI_CONFIG_PASSPHRASE`
- Minimum permissions: Select the exact stack and use read-only cloud credentials for previews.
- Credential storage: For headless runs, inject PULUMI_ACCESS_TOKEN, PULUMI_CONFIG_PASSPHRASE from the CI or platform secret manager at process start. For local interactive use, prefer the CLI or operating-system credential store when the official client supports one. Never save values in repository files.
- Never print, persist, or commit credential values.
## Allowed commands (read-only)
- `pulumi stack output --json --stack ORG/PROJECT/STACK` — R0: Reads outputs from the explicitly selected stack.
- `pulumi preview --json --stack ORG/PROJECT/STACK` — R0: Calculates infrastructure changes without applying them.
## Commands requiring explicit approval (read-only)
- None recorded.
## Forbidden commands (read-only)
- R2 `pulumi up --stack ORG/PROJECT/STACK` — Changes remote infrastructure in the selected stack.
- R3 `pulumi destroy --stack ORG/PROJECT/STACK` — Deletes managed resources from the selected stack.
## Execution rules
1. Mode boundary: R0 exact commands may be used; R1, R2, and R3 commands are forbidden.
2. Confirm the selected account, project, context, database, namespace, or environment before any command.
3. Prefer structured output using `--json`, `--diff`.
4. Capture the exact command, exit code, stdout, and stderr separately.
5. A generated prefix policy must prompt unless that exact prefix is explicitly marked suffix-safe; do not infer safety from the executable name.
6. Never broaden credentials or disable safety controls to make a command succeed.
## Official sources
- [Pulumi CLI documentation](https://www.pulumi.com/docs/iac/cli/)
- [Pulumi CLI documentation source repository](https://github.com/pulumi/pulumi)
验证记录与官方证据
CLI Finder 分开记录文档检查和真实执行。未执行过的安装、帮助、退出码与输出不能标为 Verified。