roadmap
- 01
- 02
- 03
- 04
- 05
- 06
- 07
- 08
Prepare for GitHub Actions interview questions on how workflows are triggered, run and secured. Covers workflow syntax, jobs and steps, runners, actions, secrets and environments, and caching.
32 of 32 questions in your plan: 0 known, 0 in progress, 0 due, 0 skipped, 32 new. 0 cut, outside your plan.
Questions on GitHub Actions usually arrive mid-conversation, while a candidate explains how their code reaches production, rather than in a slot set aside for them. Which team is asking decides the part of the pipeline they press on. An application developer is handed a workflow file and asked why the pull request check never ran, or why the deploy job started before the tests had finished; reading `on:`, `needs:` and `if:` correctly is most of the answer. Platform and DevOps teams ask about running Actions for a whole organization: self-hosted runners that must not carry state from one job to the next, reusable workflows that many repositories call, and a bill that grows with every row a matrix adds. Security-minded loops go straight to trust: what a pull request from a fork can read, why `pull_request_target` combined with a checkout of the fork's code is dangerous, why a third-party action is pinned to a full commit SHA rather than a tag, and how a job gets cloud credentials through OIDC instead of a stored key. Forward deployed engineers tend to meet Actions in a customer's repository more often than in their own, wiring a deploy into a pipeline someone else built, under someone else's branch rules. The [Forward Deployed Engineer roadmap](/roadmaps/forward-deployed-engineer) carries this topic inside the wider plan for that role. If the team ships through GitLab instead, the [GitLab CI/CD roadmap](/roadmaps/gitlab-ci) asks about the same ideas in that system's terms.
The [GitHub Actions primer](/topics/cloud-github-actions) already reads a workflow file from its first key to its last; this page is where you find out whether you can answer for one. Read the primer once, then drill here. A junior candidate should make [triggers](/topics/cloud-github-actions-workflow-syntax) and the [job graph](/topics/cloud-github-actions-jobs-steps) hold up first: which event starts a run, which filter quietly skips it, how `needs` orders jobs and how one job hands outputs to the next. The hub gives the same advice, because the security questions lean on both. From middle level up, [secrets and environments](/topics/cloud-github-actions-secrets-environments) belongs near the front of the queue, since that is where interviewers stop asking what a key does and start asking who can reach a credential: a fork, a third-party action, a branch that never passed an approval. Practice on a throwaway repository as you go. A workflow you have watched skip a run, or fail to see a secret on a fork's pull request, is easier to explain than one you have only read about. When Actions is one line on a longer posting, the questions worth your days sit in the lower half of the level range, where the subject is a workflow you use rather than a fleet you run, so cap level at your own band before you start.
Questions on GitHub Actions usually arrive mid-conversation, while a candidate explains how their code reaches production, rather than in a slot set aside for them. Which team is asking decides the part of the pipeline they press on. An application developer is handed a workflow file and asked why the pull request check never ran, or why the deploy job started before the tests had finished; reading on:, needs: and if: correctly is most of the answer. Platform and DevOps teams ask about running Actions for a whole organization: self-hosted runners that must not carry state from one job to the next, reusable workflows that many repositories call, and a bill that grows with every row a matrix adds. Security-minded loops go straight to trust: what a pull request from a fork can read, why pull_request_target combined with a checkout of the fork's code is dangerous, why a third-party action is pinned to a full commit SHA rather than a tag, and how a job gets cloud credentials through OIDC instead of a stored key.
Forward deployed engineers tend to meet Actions in a customer's repository more often than in their own, wiring a deploy into a pipeline someone else built, under someone else's branch rules. The Forward Deployed Engineer roadmap carries this topic inside the wider plan for that role. If the team ships through GitLab instead, the GitLab CI/CD roadmap asks about the same ideas in that system's terms.
The GitHub Actions primer already reads a workflow file from its first key to its last; this page is where you find out whether you can answer for one. Read the primer once, then drill here.
A junior candidate should make triggers and the job graph hold up first: which event starts a run, which filter quietly skips it, how needs orders jobs and how one job hands outputs to the next. The hub gives the same advice, because the security questions lean on both. From middle level up, secrets and environments belongs near the front of the queue, since that is where interviewers stop asking what a key does and start asking who can reach a credential: a fork, a third-party action, a branch that never passed an approval.
Practice on a throwaway repository as you go. A workflow you have watched skip a run, or fail to see a secret on a fork's pull request, is easier to explain than one you have only read about. When Actions is one line on a longer posting, the questions worth your days sit in the lower half of the level range, where the subject is a workflow you use rather than a fleet you run, so cap level at your own band before you start.