pipelines-azure-devops-yaml
/pipelines-azure-devops-yaml
Pomaga pisać, zmieniać i diagnozować potoki YAML w Azure DevOps, najpierw czytając nieudane uruchomienie, z wdrożeniem przez federację tożsamości.
- Autor
- JayCommit ↗
- Repozytorium
- jstack ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 04.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 2
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 04.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Pomaga pisać, zmieniać i diagnozować potoki YAML w Azure DevOps, najpierw czytając nieudane uruchomienie, z wdrożeniem przez federację tożsamości.
Oryginalny opis (EN)
Use when you write, change, or diagnose an Azure DevOps YAML pipeline. Read a failed run before you edit it, put checks on the resource, and deploy with workload identity federation.
Jak działa
- Use when
- Steps
- Safety checks
- Rollback
- Done when
Podgląd SKILL.md
---
name: pipelines-azure-devops-yaml
description: "Use when you write, change, or diagnose an Azure DevOps YAML pipeline. Read a failed run before you edit it, put checks on the resource, and deploy with workload identity federation."
license: MIT
---
# Azure DevOps YAML pipelines
Keep the pipeline in YAML and the approvals on the resource. A check that lives only inside a script is not a control.
## Use when
- A run failed or hung and you are about to edit the YAML.
- You add or edit `azure-pipelines.yml` for `example-org/example-repo`.
- You choose a Microsoft-hosted image or a self-hosted pool.
- A stage deploys to Azure or AWS, or it reads a variable group.
- You call a template from this repository or from another one.
## Steps
1. When a run has already failed or hung, open that run and read the step that exited non-zero before you edit the YAML. Follow `principle-observe-before-you-fix`. Record the run, the stage, the job and the commit. A rerun of the same revision is not the fix. GitHub Actions failures follow `pipelines-github-actions-debugging` instead of this file.
2. Declare `trigger` and `pr` for the branches this pipeline should build. In Azure Repos Git the `pr` key is ignored. Pull-request builds come from a build validation branch policy. A pull-request build is not a production deploy. Follow `principle-change-safety`.
3. Set `timeoutInMinutes` on every job. A hung step should fail inside that budget.
4. Use a Microsoft-hosted `vmImage` when the job only needs the public internet. Use a named self-hosted pool, such as a Managed DevOps Pool or a scale set pool, when the job must reach a private network. Do not send a production deploy to a pool that also builds untrusted forks. Parallel jobs beyond the free grant are billed, on either kind of pool. Follow `principle-cost-awareness`.
5. Keep secret values in a variable group linked to Key Vault, not in the YAML. Follow `azure-key-vault`. The group is a protected resource. Checks on it are configured outside this file. Follow `pipelines-environments-approvals`.
6. Log in with an Azure Resource Manager service connection that uses workload identity federation, or with the AWS federation path. Copy the issuer and the subject from that connection. Follow `pipelines-oidc-federation` and `principle-no-static-credentials`. Do not store a client secret or an access key on the connection.
7. Deploy from a deployment job in its own stage. The `environment` name is the resource that holds the checks. Those checks are not keys in this file. `azureSubscription` is the name of the federated service connection, not an id.
```yaml
stages:
- stage: production
dependsOn: test
variables:
- group: vg-example-production
jobs:
- deployment: deploy_example
timeoutInMinutes: 30
environment: production
pool:
vmImage: ubuntu-latest
strategy:
runOnce:
deploy:
steps:
- checkout: self
- task: AzureCLI@2
inputs:
azureSubscription: sc-example-production
scriptType: bash
scriptLocation: inlineScript
inlineScript: az group show --name rg-example
```
8. Insert shared steps or jobs with `template` and `parameters`. When the project requires a whole pipeline template, use `extends` and pin the repository to a tag. A branch ref moves. Follow `pipelines-reusable-templates`.
```yaml
resources:
repositories:
- repository: platform
type: git
name: example-project/platform-templates
ref: refs/tags/v1.2.0
extends:
template: templates/required.yml@platform
parameters:
environmentName: production
```
9. Give the plan connection a read role and the deploy connection a narrow deploy role. Follow `principle-least-privilege` and `azure-rbac-entra`. On AWS, follow `aws-iam-least-privilege`.
10. Promote the same build through stages. Do not rebuild the artefact in production. Follow `pipelines-artifacts-releases` when the stage ships an image.
## Safety checks
- `timeoutInMinutes` is set on every job.
- Each deploy is a `deployment` job that names an `environment`.
- Approvals, branch control and exclusive lock are on the resource, not in the steps.
- Variable groups pull secrets from Key Vault. The YAML holds none.
- The service connection is federated. It stores no static cloud key.
- A template from another repository uses `ref: refs/tags/v1.2.0`.
- The pool is Microsoft-hosted or a named self-hosted pool, and the change says which.
- A failed or hung run was read, and the failing step is named, before the YAML changed.
- Stop before an apply, a merge, or a grant. Follow `principle-stop-at-approval`.
## Rollback
Keep the previous YAML until the new pipeline has deployed once to a non-production environment. Revert the file if the new pipeline mis-deploys. Follow `principle-rollback-first`. Do not delete the environment checks to force a run through.
## Done when
- A pull-request run uses the pinned template and the named pool.
- The production stage is a deployment job whose environment has the checks in `pipelines-environments-approvals`.
- No client secret or access key was added to the pipeline or the service connection.
- A failure names the failing step and the cause. The YAML changed only because that cause required it.
Pliki w paczce (2)
- SKILL.md 5 KB
- tests/prompts.md 2 KB
Wersje
W bazie nie ma innych skilli o nazwie pipelines-azure-devops-yaml.
Migawka bazy z 13.11.2026 · wersja paczki 04c8ff88e99a (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100azure-devops-pipelines
Podaje dobre praktyki pisania plików YAML pipeline'ów Azure DevOps, aktywne dla plików azure-pipelines.yml i *.pipeline.yml. - 100azure-devops
Konfiguruje Azure Pipelines do CI/CD, potoki build i release oraz projekty Azure DevOps wraz z integracją z usługami Azure. - 100azure-pipelines
Pisze i naprawia pipeline'y YAML w Azure DevOps: etapy, szablony, cache, testy, środowiska z zatwierdzeniami i diagnozę błędów. - 100pipelines-oidc-federation
Pomaga przejść z zapisanych kluczy i sekretów w potokach na federację OIDC z Azure lub AWS, z dokładnym tematem tożsamości. - 100ci-cd
Konfiguruje CI/CD dla aplikacji .NET w GitHub Actions i Azure DevOps YAML: budowanie, testy, publikacja, wdrożenie oraz wysyłka pakietów NuGet. - 100azure-devops-workflow
Pomaga w Azure DevOps: triażuje work items, przegląda pull requesty i diagnozuje błędy buildów oraz pipeline'ów.