Azure CLI logicapp extension을 사용하여 Azure Logic Apps Standard 프로젝트의 작성, 테스트 및 배포를 관리합니다.
- GitHub 스타
- 41,038
- 파일
- 7
- 마지막 커밋
- 2026년 11월 11일
- 라이선스
- MIT
- 허용된 도구
- —
azure-logic-apps-standard-cli skill은 무엇을 하나요?
azure-logic-apps-standard-cli은(는) agent skill입니다. Azure CLI logicapp extension을 사용하여 Azure Logic Apps Standard 프로젝트의 작성, 테스트 및 배포를 관리합니다.
skill의 SKILL.md에서 생성된 요약입니다.
azure-logic-apps-standard-cli skill 정보
Azure Logic Apps Standard 프로젝트를 관리하는 개발자는 Azure CLI logicapp extension을 통해 워크플로의 전체 수명 주기를 처리할 수 있습니다. 이 SKILL은 로컬 개발과 클라우드 운영에 특화되어 Standard 워크플로의 작성, 검증, 테스트 및 배포를 위한 통합 인터페이스를 제공합니다. 사용자가 새로운 프로젝트를 스캐폴딩하거나, 커넥터를 탐색하거나, ID 기반 연결을 관리하거나, 네이티브 워크플로 테스트를 수행해야 할 때 에이전트가 이 SKILL을 로드해야 합니다.
이 SKILL은 Azure CLI 2.67.0 이상 버전과 Python 3.10 런타임이 필요합니다. 패키징된 배포 파일, 검증된 워크플로 구조 및 관리형 연결 구성을 생성합니다. 일반적인 Azure 관리 도구와 달리 이 SKILL은 Standard 티어에 특화되어 있으며, 호스트 시작 및 중지 같은 로컬 런타임 관리와 managed identity를 기본으로 사용하는 ID 기반 연결 관리와 같은 특정 절차를 포함합니다. Consumption 전용 워크플로 개발용으로는 설계되지 않았습니다.
agent가 수행할 수 있는 작업
- 1새 프로젝트 스캐폴딩로컬에 새로운 Azure Logic Apps Standard 프로젝트 구조를 생성합니다.
- 2managed identity 구성보안 리소스 액세스를 위한 ID 기반 연결을 설정합니다.
skill의 SKILL.md를 기반으로 작성되었습니다. 모든 사용 사례는 해당 문서의 인용구와 대조 확인되었습니다.
azure-logic-apps-standard-cli skill 설치 방법
각 명령은 skill 폴더를 클라이언트의 skills 디렉터리에 복사합니다. 클라이언트를 재시작하거나 새 세션을 시작하여 로드하세요.Claude Code에 azure-logic-apps-standard-cli skill 설치
~/.claude/skills/azure-logic-apps-standard-cli/SKILL.md
$ npx degit github/awesome-copilot/skills/azure-logic-apps-standard-cli ~/.claude/skills/azure-logic-apps-standard-cliClaude Code (project)에 azure-logic-apps-standard-cli skill 설치
.claude/skills/azure-logic-apps-standard-cli/SKILL.md
$ npx degit github/awesome-copilot/skills/azure-logic-apps-standard-cli .claude/skills/azure-logic-apps-standard-cliCodex에 azure-logic-apps-standard-cli skill 설치
~/.agents/skills/azure-logic-apps-standard-cli/SKILL.md
$ npx degit github/awesome-copilot/skills/azure-logic-apps-standard-cli ~/.agents/skills/azure-logic-apps-standard-cliGemini CLI에 azure-logic-apps-standard-cli skill 설치
~/.gemini/skills/azure-logic-apps-standard-cli/SKILL.md
$ npx degit github/awesome-copilot/skills/azure-logic-apps-standard-cli ~/.gemini/skills/azure-logic-apps-standard-cliClaude.ai에 azure-logic-apps-standard-cli skill 설치
npx degit github/awesome-copilot/skills/azure-logic-apps-standard-cli azure-logic-apps-standard-cli && zip -r azure-logic-apps-standard-cli.zip azure-logic-apps-standard-cli # claude.ai → Customize → Skills → + → Create skill → Upload a skill → azure-logic-apps-standard-cli.zip
자동 검사
2026년 11월 15일에 실행됨전체 보고서 →자동 검사를 통과했습니다. 보안 감사는 아닙니다. 에이전트가 실행하기 전에 스크립트를 읽어 보세요. 검사 방식
- 소스 저장소 연결 및 접근 가능github/awesome-copilot/skills/azure-logic-apps-standard-cli is public and reachablegithub.com/github/awesome-copilot/tree/HEAD/skills/azure-log ↗통과
- 오픈 소스 라이선스 (OSI)MITspdx.org/licenses/MIT.html ↗통과
- 최근 1년 내 커밋 또는 릴리스Last commit 2026-11-11github.com/github/awesome-copilot/commits ↗통과
- 스크립트에 다운로드 후 실행이나 자격 증명 탈취가 없음No scripts in the skill folder통과
- SKILL.md에 숨겨진 지침 없음No override phrases, safeguard switches, credential requests or hidden charactersgithub.com/github/awesome-copilot/blob/HEAD/skills/azure-log ↗통과
- 사전 허용 도구가 제한됨No pre-approved tools; the agent asks before using onegithub.com/github/awesome-copilot/blob/HEAD/skills/azure-log ↗통과
- SKILL.md가 Agent Skills 형식을 따름Valid name and description; says when to use it; 293 linesgithub.com/github/awesome-copilot/blob/HEAD/skills/azure-log ↗통과
SKILL.md
GitHub에서 보기Logic Apps Standard CLI
Use az logicapp as the primary interface for supported Standard project, workflow,
connection, validation, runtime, diagnostic, testing, and packaging operations.
Do not manually recreate files or Azure wiring owned by a supported command.
Manual workflow authoring, workspace files, ignore files, and infrastructure-as-code
remain appropriate where the CLI does not own that operation.
Reviewed baseline: extension 1.0.0b1. It is a public preview for evaluation and development, not a claim of production readiness. Use non-production resources for evaluation. Existing core app-management, app-settings, and ZIP-deployment commands are distinct from the extension's preview commands, even where reference pages label extension commands GA. Declared platform support is not live-runtime qualification.
Mandatory procedure routing
This folder is one installable skill package. Resolve bundled files relative to this
SKILL.md, not the customer's working directory. Install/copy the entire folder.
These references are required procedures, not optional background reading.
See the package README for installation, provenance, and maintenance notes.
Before the corresponding operation, read every applicable reference below. For combined tasks, read each required procedure before its phase. If a required file is missing or unreadable, report that operation as blocked. Do not invent syntax, bypass its prerequisites, or fetch an unreviewed substitute.
| Requested operation | Required bundled procedure |
|---|---|
| Install/check dependencies; diagnose, start, or stop a host | Local runtime |
| Discover connectors; create/change/test/delete connections; configure identity or ACLs | Connections and identity |
| Inspect/export/change live workflows; invoke/diagnose/cancel/replay runs | Live workflows and runs |
| Scaffold, plan, run, or inspect native workflow tests | Workflow testing |
| Package, deploy, or change deployment exclusions | Packaging and deployment |
Basic local scaffolding, workspace creation, and offline validation are specified here. Do not start a host, restore packages, invoke workflows, or mutate Azure merely because the skill was activated. Select only the procedures needed for the requested task.
Non-negotiable rules
- Inspect before mutation. Establish the actual project root and requested scope. Inspect affected workflows, configuration structure/references, ignore files, workspaces, and uncommitted changes. Do not dump secret setting values.
- Preserve existing conventions. Do not restructure repositories, convert workflow kinds, migrate authentication, normalize aliases, or replace user files unsolicited.
- Use managed identity by default. For new connections and deployed app configuration, use another authentication method only when explicitly requested by the customer or unsupported by the selected connector/operation. Authentication failure is not proof of lack of support. Never silently fall back to secrets.
- Enable identity as part of new-project delivery. Configure the supported local identity-based connection mechanism. For deployment-facing work, enable app identity through the authorized provisioning/deployment process. Preserve an existing identity convention; otherwise default to system-assigned identity. Local scaffolding does not authorize Azure mutation; report unapproved cloud prerequisites explicitly.
- Verify distinct principals and permissions. A local process does not acquire a
deployed app's managed identity. Follow the connector's documented local mechanism;
do not assume every connector uses
DefaultAzureCredential. Verify local and cloud principals, connection access, and least-privilege data-plane roles separately. - Discover, do not guess. Obtain connector IDs, operation IDs, parameter sets, and required input shapes from the actual runtime or regional managed catalog. Verify authentication support using metadata and authoritative connector documentation.
- Protect secrets and sensitive outputs. Never put secret values in command arguments, source control, examples, or shared logs. Treat existing definitions, exports, and provider output as potentially sensitive; protect and review them before sharing. Use supported stdin secret input and setting references. Required protected local/runtime storage is permitted; local settings and test artifacts must be ignored and excluded from deployment.
- Keep environments configurable. Put deployment-specific endpoints, resource IDs, and environment values in settings or parameters, not hardcoded workflow literals.
- Guard replacements and concurrency. Inspect before every create; treat creation
as non-idempotent. Use
--forceonly for reviewed replacement. Export live workflows before replacement/deletion, retain revisions, and obtain approval for non-atomic writes. Do not bypass conflicts or blindly retry uncertain mutations. - Authorize effects, not just targets. Starting local triggers and running test/build code can contact real systems. Establish permitted effects first. Obtain explicit authorization for Azure mutations, cancellation, and replay; preview is not a sandbox.
- Keep evidence honest. Read structured results even on nonzero exits. Distinguish configured, authorized, accepted, observed, and verified states. An open port, accepted invocation, generated plan, or inspected report is not successful workflow execution.
- Validate without manufacturing success. Validate incrementally; require strict full-project validation before release packaging/deployment. Preserve stricter repository policy. Do not remove warnings, assertions, outputs, or guards merely to obtain a passing result.
- Own cleanup precisely. Stop only task-owned runtimes unless asked to leave them running; verify shutdown and shared dependency ownership. Never kill processes by name or stop another user's host/emulator.
- Preserve deployment boundaries. Merge ignore rules without replacing unrelated entries; inspect the actual package. Do not deploy local settings, tests, state, editor files, logs, settings templates, or generated packages.
- Keep connector semantics specialized. Use available connector-specific guidance for delivery guarantees, poison handling, and operation-specific tests. If unavailable, use authoritative connector documentation; do not invent specialized behavior.
Defaults and consequential choices
| Decision | Rule |
|---|---|
| Workflow kind | Preserve existing kinds; new workflows default to Stateful |
| Stateful vs Stateless | Stateful for durable history/long-running work; choose Stateless deliberately for suitable short-lived, low-latency work |
| Initial scope | Local authoring, unless the request includes further operations |
| Authentication | Managed identity, subject only to the exceptions above |
| Identity type | Existing convention, otherwise system-assigned; user-assigned for explicit stable/shared/pre-provisioned identity requirements |
| Project root/layout | Existing repository convention; otherwise the smallest suitable layout |
| Workspace | Optional; reuse an existing workspace and ask only if intent is unresolved |
| Context | Optional saved shortcut; never change active selection without authorization |
| Automation | Explicit project path and runtime/cloud target coordinates |
| Runtime lifetime | Stop task-owned hosts after checks unless explicitly asked to retain them |
Infer paths and established choices from the request/repository. Ask only about missing, consequential decisions. Do not insist on three scaffold questions when answers are known.
1. Verify tooling and select the target
Inspect first; these commands do not install the extension:
az version
az extension list --query "[?name=='logicapp'].{name:name,version:version}" --output table
Minimums for the reviewed release: Azure CLI 2.67.0 and its Python runtime 3.10. If the extension is absent and installation is authorized, use the public extension index:
az extension add --name logicapp --version 1.0.0b1 --allow-preview true
az extension show --name logicapp --output json
az logicapp --help
Check the exit after each command; do not continue after failed installation.
If the requested version is unavailable, report the index/network/version error.
Do not substitute a private wheel, disable TLS, or bypass organizational package policy.
Do not automatically upgrade an installed extension. For a different installed version,
inspect each affected command's --help and current official contracts before execution;
report material differences instead of assuming this baseline still applies.
Detect OS and shell before using examples. Single-line azurecli blocks avoid shell-specific
continuations. App, Orders, and uppercase identifiers are illustrative names: substitute
the inspected target, quote paths containing spaces, and use the host OS's path separator.
PowerShell/native-wrapper quoting for expressions is specified in the connections procedure.
| Operation | Target contract |
|---|---|
| Local file authoring/validation | Explicit --path; some commands also accept a local context |
| Runtime API inspection/execution | --local --port PORT, or cloud --resource-group GROUP --name APP |
| Runtime lifecycle | Local runtime start/show/stop flags; not cloud host management |
| Runtime doctor | Local --path and optional --port, or local --context; no --local |
| Managed connection creation | --name CONNECTION and --logic-app-name APP, not interchangeable |
| Live workflow enable/disable | Deployed apps only |
For API-targeting commands, explicit --port requires --local. A named context owns its
coordinates: do not mix it with explicit API target flags. Inspect command-specific help
for file-command context combinations; do not assume one universal precedence rule.
Cloud work requires the intended Azure login, tenant, subscription, and permissions.
Verify account selection independently: saved contexts do not bind subscription/tenant.
Avoid changing global account selection; prefer explicit --subscription in automation.
2. Inspect and scaffold locally
Classify the destination as new, an existing Standard project, or a nonempty unrecognized
directory. The project/host root contains host.json; workflow directories are its direct
children. The repository root may itself be that project root.
For an existing project, inspect host.json, workflow files, connections.json,
parameters.json when present, settings names/references, ignore files, workspaces, and
affected uncommitted changes. Preserve unrelated content and avoid exposing local values.
For a new project, choose a minimal project-at-root layout or a solution/monorepo child
project according to requirements. Support directories such as docs, requests, fixtures,
tests, logs, and artifacts are optional and created only for actual work. Prefer package
output outside the host root. Do not author .azurelogicapps runtime state.
After inspecting the destination, scaffold using explicit paths:
az logicapp project create --path App --workflow Orders
Add --workflow-kind Stateless only for an intentional Stateless choice.
Use --adopt only for a reviewed nonempty unrecognized directory, not as routine setup.
Inspect generated host/settings/connection/workflow files before extending them.
Create additional workflows without replacing existing ones:
az logicapp workflow create --path App --workflow Notify --kind Stateful
--file FILE accepts a reviewed full workflow file or bare definition for local import.
Creation is local-only and non-idempotent. Do not use it to add a cloud workflow.
Before writing sensitive local settings, confirm effective Git exclusions. Merge rules
relative to the actual repository and host roots; never copy .gitignore over .funcignore.
Always exclude local settings/state from source control and deployment. If changing
deployment exclusions, first read the packaging procedure and follow its concrete rules.
Gate: Correct project root, requested workflow kinds, preserved user files, effective secret exclusions, and identity-based configuration or an explicitly reported prerequisite.
3. Optional workspace and context
A .code-workspace file is optional; opening the host directory is valid. Reuse an existing
workspace. When a new multi-root workspace is requested, use relative paths to real folders
and put the Logic Apps host project first. Do not create empty folders just to populate it.
Adapt this JSON to the selected layout; it is not a required filename or folder scheme:
{
"folders": [
{ "name": "Logic App", "path": "App" }
]
}
Preserve existing workspace settings and names. Add docs/test roots only when useful and
present. For a project-at-root workspace, the project path is ".".
Saved contexts are optional. Inspect context list before choosing a unique project-specific
name. If requested, create a shortcut without implicitly selecting it:
az logicapp context list --output table
az logicapp context create --name orders-local --path App --port 7071
az logicapp context show --name orders-local
Use context use --name NAME or creation's --use only when active selection is authorized.
context update preserves omitted fields and does not select a target; editing an active
target prompts unless --yes is authorized. A kind change requires explicit local path/port
or cloud resource group/app coordinates. Inspect its help before changing kind.
context delete deletes only a shortcut; deleting the active one clears selection without
choosing a replacement. Stores follow AZURE_CONFIG_DIR; no credentials belong in them.
4. Author and validate
Preserve workflow structure/kind. Use stable action/trigger names. Implement the required ordering, branching, retries, timeouts, and failure paths; terminal success/failure paths must not both execute. Use discovered operation shapes and configurable environment values. Create representative non-sensitive fixtures only when needed.
For built-in connector discovery, read both the local-runtime and connections procedures before establishing safe host startup and querying the catalog. For runtime-backed validation, read the local-runtime procedure and establish readiness first. Native tests can instead use authored mocks without starting a host.
Validate the workflow under active development:
az logicapp workflow validate --path App --workflow Orders
For completed project authoring and before release packaging/deployment:
az logicapp workflow validate --path App --all --fail-on-warning
When runtime-backed validation is required, first establish host readiness:
az logicapp workflow validate --path App --all --fail-on-warning --with-runtime --local --port 7071
Do not start a host solely for an offline edit/review. Assess findings and preserve repository policy; report unrelated pre-existing failures separately without weakening release gates. Static validation, runtime-template validation, native assertions, and live integration checks are separate evidence. A completed runtime validation response is required; accepted asynchronous validation alone does not establish completion.
Completion evidence
Apply only gates relevant to the requested task; do not turn an inspection into a deployment.
| Task | Required evidence |
|---|---|
| Scaffold/author | Requested files/kinds and behavior; affected validation; no unsolicited restructuring |
| Connections/identity | Discovered shape, setting references, selected principal, verified grants or explicit blockers; no secrets exposed |
| Full-project release | Strict full-project validation passes; required runtime validation completes |
| Native tests | Reviewed meaningful assertions, actual executed counts/outcome, skipped counts and limitations |
| Live execution/diagnosis | Actual supported run evidence, required outputs/effects, failed action/platform error where available |
| Live mutation | Retained export/revision, guarded write result, separate observed runtime behavior |
| Cancellation/replay | Requested/accepted/observed/unknown distinguished; no inferred new run IDs |
| Package/deploy | Inspected archive/exclusions; authorized prerequisites; required post-deploy verification |
| Runtime ownership | Task-owned processes stopped and verified, or explicitly retained with coordinates/ownership |
State concrete missing permissions, dependencies, resources, unexecuted checks, retained resources, and uncertain effects. Do not call configured identity authorized, an accepted request complete, a skipped suite passing, or a blocked prerequisite satisfied.
이 스킬의 파일
7azure-logic-apps-standard-cli skill 질문
Claude Code에 azure-logic-apps-standard-cli skill을 어떻게 설치하나요?
npx degit github/awesome-copilot/skills/azure-logic-apps-standard-cli ~/.claude/skills/azure-logic-apps-standard-cli를 실행합니다. Claude Code는 ~/.claude/skills(모든 프로젝트) 또는 .claude/skills(한 프로젝트)에서 skills를 로드하며, 설명이 작업과 일치할 때 skill을 사용합니다.
azure-logic-apps-standard-cli skill을 사용할 수 있는 클라이언트는 무엇인가요?
Agent Skills 형식을 읽는 모든 클라이언트(Claude Code, Codex, Gemini CLI, claude.ai(ZIP으로 업로드))가 사용할 수 있습니다. skill 내 스크립트는 클라이언트가 코드 실행을 허용하는 경우에만 실행됩니다.
이 SKILL을 사용하기 위한 최소 요구 사항은 무엇인가요?
logicapp extension이 설치된 Azure CLI 2.67.0 이상 버전과 Python 3.10 이상의 런타임이 필요합니다.
Consumption 전용 워크플로에도 사용할 수 있나요?
아니요, 이 SKILL은 Standard 워크플로 개발 전용이며 Consumption 전용 워크플로를 지원하지 않습니다.
새로운 연결에 대한 인증은 어떻게 처리해야 하나요?
다른 방법이 명시적으로 요청되지 않는 한, 새로운 연결 및 배포된 앱 구성에는 기본적으로 managed identity를 사용해야 합니다.