Motivation
Link to the gh-aw PR: github/gh-aw#47874 — Fix dangling secret_verification_result reference in conclusion condition when a workflow configures a top-level environment:.
When a gh-aw .md source file includes a top-level environment: field, the compiled lock file's activation job skips the validate-secret step, meaning secret_verification_result is never emitted as an output. Before this fix, buildConclusionJobCondition unconditionally referenced that output, causing actionlint failures on the generated lock file. No existing E2E test exercises the environment: frontmatter field at all.
Proposed test
- Workflow file:
test-copilot-environment-field.md
- Trigger:
workflow_dispatch
- Engine: copilot
- Safe output:
create-issue (simple baseline output to confirm the workflow runs end-to-end)
- Variant: standard
Minimal test prompt sketch
A simple workflow with a top-level environment: test (or an existing named environment in the repo) in the frontmatter. The agent creates an issue with a fixed title. The primary assertion is that the compiled lock file passes actionlint and the workflow completes successfully — validating that the environment: field does not break the conclusion job condition.
New fixtures or secrets needed
A named GitHub Actions environment (e.g. test) must exist in githubnext/gh-aw-test. This is a one-time repo setup step.
Notes
This is a regression test for the actionlint-breaking bug fixed in PR #47874. It exercises a frontmatter field currently absent from all 100+ existing test sources. The environment: field is orthogonal to the engine, so the copilot variant alone is sufficient initially.
Generated by 🔍 Suggest New E2E Tests · sonnet46 · 32.7 AIC · ⌖ 8.96 AIC · ⊞ 6.1K · ◷
Motivation
Link to the gh-aw PR: github/gh-aw#47874 — Fix dangling
secret_verification_resultreference in conclusion condition when a workflow configures a top-levelenvironment:.When a gh-aw
.mdsource file includes a top-levelenvironment:field, the compiled lock file'sactivationjob skips thevalidate-secretstep, meaningsecret_verification_resultis never emitted as an output. Before this fix,buildConclusionJobConditionunconditionally referenced that output, causingactionlintfailures on the generated lock file. No existing E2E test exercises theenvironment:frontmatter field at all.Proposed test
test-copilot-environment-field.mdworkflow_dispatchcreate-issue(simple baseline output to confirm the workflow runs end-to-end)Minimal test prompt sketch
A simple workflow with a top-level
environment: test(or an existing named environment in the repo) in the frontmatter. The agent creates an issue with a fixed title. The primary assertion is that the compiled lock file passesactionlintand the workflow completes successfully — validating that theenvironment:field does not break the conclusion job condition.New fixtures or secrets needed
A named GitHub Actions environment (e.g.
test) must exist ingithubnext/gh-aw-test. This is a one-time repo setup step.Notes
This is a regression test for the
actionlint-breaking bug fixed in PR #47874. It exercises a frontmatter field currently absent from all 100+ existing test sources. Theenvironment:field is orthogonal to the engine, so the copilot variant alone is sufficient initially.