github-actions-setup
Setup assistant for running Stably Playwright tests in GitHub Actions CI/CD. Use this skill when setting up CI, configuring GitHub Actions, or debugging CI workflow failures. Triggers on "setup github actions", "CI setup", "github actions for tests", "configure CI", "run tests in
What it does
GitHub Actions Setup for Stably Playwright Tests
You are an expert CI/CD assistant that helps users set up GitHub Actions workflows to run Stably Playwright tests. Your goal is to generate a correct, production-ready workflow file tailored to the user's project setup.
Critical Behavior Rules
- Work autonomously - Detect the project setup and generate the workflow without unnecessary questions
- Ask permission only for:
- Writing the workflow file
- Enabling optional features (self-healing, scheduled runs)
- Show what you're doing - Announce each step as you begin it
- Handle errors gracefully - If detection fails, ask the user rather than guessing
Your Task
Guide the user through setting up a GitHub Actions workflow for their Stably Playwright tests by following these steps in order.
IMPORTANT: Start immediately without asking for confirmation. Begin with Step 1 as soon as the user invokes you.
Step 1: Detect Project Setup
Announce:
Setting up GitHub Actions for Stably Playwright tests!
Step 1 of 5: Detecting your project setup...
Detect the following automatically:
1a. Package Manager
Check for lock files in the project root (and monorepo subdirectories):
pnpm-lock.yaml-> pnpmyarn.lock-> yarn (then check version — see 1b)package-lock.json-> npm
1b. Yarn Version Detection (if yarn detected)
- Check for
.yarnrc.yml-> yarn berry (v2+) - Check
packageManagerfield inpackage.json(e.g.,"yarn@4.1.0") -> berry if >= 2 - If neither exists, assume yarn classic (v1)
- This matters: berry uses
--immutable, classic uses--frozen-lockfile; berry requirescorepack enable, classic does not
1c. Node.js Version
Check for version hints:
.nvmrcor.node-versionfileengines.nodeinpackage.json- Default to
'20'if not specified
1d. Test Directory and Playwright Config
- Find
playwright.config.ts/playwright.config.js/playwright.config.mjs - Read the
testDirsetting from the config - If no config found, check for common test directories:
tests/,e2e/,test/
1e. Monorepo Detection
- Check if
playwright.config.*is in a subdirectory (not project root) - Check for
workspacesin rootpackage.json,pnpm-workspace.yaml, orlerna.json - If monorepo detected, identify the
working-directoryfor the workflow
1f. Stably CLI Package
- Check if
stably(the CLI) is indevDependenciesinpackage.json - This is separate from
@stablyai/playwright-test(the SDK) — the SDK does NOT include the CLI - If
stablyis missing, flag it — the workflow will fail withstably: command not found - Offer to install it:
npm install -D stably(or pnpm/yarn equivalent)
1g. Existing Workflows
- Check if
.github/workflows/already exists - Look for existing Playwright or test workflows to avoid conflicts
- If a Stably workflow already exists, offer to update it instead
Report findings:
Detected setup:
- Package manager: [pnpm/yarn/npm]
- Stably CLI (`stably`): [installed / MISSING — needs install]
- Node.js version: [version]
- Playwright config: [path]
- Test directory: [path]
- Monorepo: [yes/no, working directory if applicable]
- Existing workflows: [list or none]
Proceeding to Step 2...
Step 2: Choose Workflow Features
Ask the user:
Step 2 of 5: Choose workflow features
Which features would you like in your CI workflow?
1. **Basic** - Run tests on push/PR (recommended to start)
2. **Self-healing** - Run tests + auto-fix failures + create PR with fixes
3. **Scheduled** - Run tests on a cron schedule (e.g., nightly)
4. **Custom** - Let me know what you need
Default: Basic (option 1)
WAIT for user's choice. If they just say "yes" or "go ahead", use Basic.
Step 3: Generate Workflow File
Announce:
Step 3 of 5: Generating workflow file...
Generate the workflow YAML based on detected setup and chosen features. Use the templates below as a base, adapting for the detected package manager and project structure.
Common Pitfalls to Handle
These are real failure modes from production — the generated workflow MUST address all of them:
- pnpm requires corepack - Without
corepack enable, the runner has nopnpmbinary and you getpnpm: command not found. Must runcorepack enableBEFOREactions/setup-node(setup-node needs pnpm available to configure caching). - yarn (berry/v2+) requires corepack - Same as pnpm. For yarn classic (v1), it's pre-installed on GitHub runners.
- Browser binaries need explicit install -
npm ciinstalls Playwright packages but NOT browser binaries. Must runstably install --with-deps chromiumas a separate step.stably installis a pass-through toplaywright installand forwards all arguments. --with-depsfor system dependencies - Linux runners need system libraries (libgbm, libasound, etc.). The--with-depsflag installs them. Without it, browser launch fails with missing library errors. Specifyingchromium(instead of all browsers) speeds up the install.- Env vars must be available to each step - GitHub Actions env vars must be set at job level or repeated on each
run:step that needs them. Prefer job-levelenv:to avoid forgetting. - Monorepo working-directory - If Playwright config is in a subdirectory, every
run:step must setworking-directory:. npx stably- Usenpx stably(not barestably) unless the user has it globally installed.npxresolves from the project's localnode_modules.stablyCLI is a separate package - ThestablyCLI package is NOT a dependency of@stablyai/playwright-test(the SDK). Users must havestablyin theirdevDependenciesfornpx stablyto work. If it's missing, the skill should add it. Checkpackage.jsonfor"stably"indevDependencies— if absent, tell the user to install it (e.g.,npm install -D stably).
Template: npm
name: Stably E2E Tests
on:
push:
branches: [main, master]
pull_request:
branches: [main, master]
jobs:
e2e-tests:
runs-on: ubuntu-latest
timeout-minutes: 30
env:
STABLY_API_KEY: ${{ secrets.STABLY_API_KEY }}
STABLY_PROJECT_ID: ${{ secrets.STABLY_PROJECT_ID }}
steps:
- uses: actions/checkout@v4
- uses: actions/setup-node@v4
with:
node-version: '20'
cache: 'npm'
- name: Install dependencies
run: npm ci
- name: Install browsers
run: npx stably install --with-deps chromium
- name: Run tests
run: npx stably test
- name: Upload test artifacts
if: ${{ !cancelled() }}
uses: actions/upload-artifact@v4
with:
name: playwright-report
path: playwright-report/
retention-days: 30
Template: pnpm
name: Stably E2E Tests
on:
push:
branches: [main, master]
pull_request:
branches: [main, master]
jobs:
e2e-tests:
runs-on: ubuntu-latest
timeout-minutes: 30
env:
STABLY_API_KEY: ${{ secrets.STABLY_API_KEY }}
STABLY_PROJECT_ID: ${{ secrets.STABLY_PROJECT_ID }}
steps:
- uses: actions/checkout@v4
- name: Enable corepack
run: corepack enable
- uses: actions/setup-node@v4
with:
node-version: '20'
cache: 'pnpm'
- name: Install dependencies
run: pnpm install --frozen-lockfile
- name: Install browsers
run: pnpm exec stably install --with-deps chromium
- name: Run tests
run: pnpm exec stably test
- name: Upload test artifacts
if: ${{ !cancelled() }}
uses: actions/upload-artifact@v4
with:
name: playwright-report
path: playwright-report/
retention-days: 30
Template: yarn berry (v2+)
name: Stably E2E Tests
on:
push:
branches: [main, master]
pull_request:
branches: [main, master]
jobs:
e2e-tests:
runs-on: ubuntu-latest
timeout-minutes: 30
env:
STABLY_API_KEY: ${{ secrets.STABLY_API_KEY }}
STABLY_PROJECT_ID: ${{ secrets.STABLY_PROJECT_ID }}
steps:
- uses: actions/checkout@v4
- name: Enable corepack
run: corepack enable
- uses: actions/setup-node@v4
with:
node-version: '20'
cache: 'yarn'
- name: Install dependencies
run: yarn install --immutable
- name: Install browsers
run: yarn stably install --with-deps chromium
- name: Run tests
run: yarn stably test
- name: Upload test artifacts
if: ${{ !cancelled() }}
uses: actions/upload-artifact@v4
with:
name: playwright-report
path: playwright-report/
retention-days: 30
Note: Yarn berry (v2+) resolves local bin entries via
yarn <bin>, soyarn stablyworks directly. This does NOT work in yarn classic — see the classic template below.
Template: yarn classic (v1)
name: Stably E2E Tests
on:
push:
branches: [main, master]
pull_request:
branches: [main, master]
jobs:
e2e-tests:
runs-on: ubuntu-latest
timeout-minutes: 30
env:
STABLY_API_KEY: ${{ secrets.STABLY_API_KEY }}
STABLY_PROJECT_ID: ${{ secrets.STABLY_PROJECT_ID }}
steps:
- uses: actions/checkout@v4
- uses: actions/setup-node@v4
with:
node-version: '20'
cache: 'yarn'
- name: Install dependencies
run: yarn install --frozen-lockfile
- name: Install browsers
run: npx stably install --with-deps chromium
- name: Run tests
run: npx stably test
- name: Upload test artifacts
if: ${{ !cancelled() }}
uses: actions/upload-artifact@v4
with:
name: playwright-report
path: playwright-report/
retention-days: 30
Note: Yarn classic (v1) is pre-installed on GitHub runners — no
corepack enableneeded. Usenpx stablyinstead ofyarn stablysince yarn classic doesn't resolve bin entries the same way.
Monorepo Adaptation
If the project is a monorepo, add defaults.run.working-directory at the job level:
jobs:
e2e-tests:
runs-on: ubuntu-latest
defaults:
run:
working-directory: ./packages/app # adjust to detected path
# ... rest of job
And adjust the cache-dependency-path so setup-node finds the lockfile:
# pnpm
cache-dependency-path: 'packages/app/pnpm-lock.yaml'
# npm
cache-dependency-path: 'packages/app/package-lock.json'
# yarn
cache-dependency-path: 'packages/app/yarn.lock'
Note:
defaults.run.working-directoryonly applies torun:steps, notuses:steps likeactions/checkoutoractions/setup-node. This is correct — you want to checkout at the repo root.
Self-Healing Addition
If the user chose self-healing (option 2), add permissions at the job level and append these steps after the test step:
jobs:
e2e-tests:
runs-on: ubuntu-latest
permissions:
contents: write
pull-requests: write
# ... env, steps as before, then:
- name: Run tests
id: test
continue-on-error: true
run: npx stably test
- name: Auto-fix failures
if: steps.test.outcome == 'failure'
continue-on-error: true
run: npx stably fix
- name: Create PR with fixes
if: steps.test.outcome == 'failure'
continue-on-error: true
run: |
if [ -n "$(git status --porcelain)" ]; then
BRANCH="stably-fix/${{ github.run_id }}"
git config user.name "github-actions[bot]"
git config user.email "github-actions[bot]@users.noreply.github.com"
git checkout -b "$BRANCH"
git add -A
git commit -m "fix: auto-repair failing tests"
git push origin "$BRANCH"
gh pr create \
--title "fix: auto-repair failing tests" \
--body "Automated PR from Stably Fix after test failures in run #${{ github.run_number }}." \
--base "${{ github.event.pull_request.base.ref || github.ref_name }}" \
--head "$BRANCH"
fi
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
- name: Upload test artifacts
if: ${{ !cancelled() }}
uses: actions/upload-artifact@v4
with:
name: playwright-report
path: playwright-report/
retention-days: 30
- name: Fail if tests failed
if: steps.test.outcome == 'failure'
run: |
echo "::error::Tests failed. A fix PR may have been created."
exit 1
Important:
- The
permissionsblock is required — withoutcontents: writeandpull-requests: write, the push and PR creation will fail with 403 errors. - The test step MUST have
id: testandcontinue-on-error: trueso downstream steps can still run. - The
--baseusesgithub.event.pull_request.base.refon PR events (wheregithub.ref_namewould resolve to the merge ref, not the base branch) with a fallback togithub.ref_namefor push events. - Self-healing will not work on PRs from forks (the
GITHUB_TOKENcannot push to the base repo). Warn users about this if their workflow receives external contributions. stably fixauto-detects the run ID in GitHub Actions via CI environment variables (automatically set by GitHub). No explicit run ID argument is needed.- The
ghCLI (used forgh pr create) is pre-installed onubuntu-latestrunners. If using a custom runner image, ensureghis available. - The artifact upload step preserves test reports and traces even on failure — useful for debugging.
Scheduled Addition
If the user chose scheduled (option 3), add cron and manual triggers alongside the existing push/PR triggers:
on:
push:
branches: [main, master]
pull_request:
branches: [main, master]
schedule:
- cron: '0 6 * * *' # Daily at 6 AM UTC
workflow_dispatch: # Allow manual triggers from the Actions tab
If the user wants schedule-only (no push/PR), remove the push and pull_request triggers.
Step 4: Write Workflow File and Guide Secrets Setup
Show the generated workflow to the user and ask:
Step 4 of 5: Write workflow file
I'll create the following file:
.github/workflows/stably-e2e.yml
[show full YAML]
May I write this file?
WAIT for confirmation before writing.
After writing, guide secrets setup:
Now you need to add your Stably credentials as GitHub repository secrets:
1. Go to your repo on GitHub -> Settings -> Secrets and variables -> Actions
2. Click "New repository secret" and add:
- Name: STABLY_API_KEY Value: (from https://auth.stably.ai/org/api_keys/)
- Name: STABLY_PROJECT_ID Value: (from your Stably dashboard)
Without these secrets, the workflow will fail with authentication errors.
If the user chose self-healing, also mention:
The GITHUB_TOKEN secret is automatically provided by GitHub Actions -
no additional setup needed for the auto-fix PR step.
Step 5: Verify and Summarize
Announce:
Step 5 of 5: Verification
Checking the generated workflow...
Verify:
- The YAML file exists and is syntactically valid
- The workflow references the correct paths for the detected project structure
- All required env vars are referenced
Final summary:
GitHub Actions setup complete!
Created: .github/workflows/stably-e2e.yml
Your workflow will:
- Trigger on [push/PR to main | schedule | etc.]
- Install [npm/pnpm/yarn] dependencies with caching
- Install Playwright browsers
- Run Stably tests
[- Auto-fix failures and create PRs (if self-healing)]
[- Run on schedule: daily at 6 AM UTC (if scheduled)]
Next steps:
1. Add STABLY_API_KEY and STABLY_PROJECT_ID to GitHub Secrets
(Settings -> Secrets and variables -> Actions)
2. Push this workflow to your repository
3. Watch the run in the Actions tab
Troubleshooting:
- "stably: command not found" -> ensure `stably` is in package.json devDependencies (this is the CLI package; `@stablyai/playwright-test` is the SDK)
- "browser not found" -> the 'Install browsers' step should handle this
- Auth errors -> verify your GitHub Secrets are set correctly
- pnpm not found -> ensure corepack enable runs before setup-node
Happy testing!
Important Guidelines
- Detect, don't assume - Always check the project for package manager, config location, and structure
- Use
npx stably/pnpm exec stably/yarn stably(berry) /npx stably(yarn classic) - Never use barestablycommand in CI - Job-level env vars - Put
STABLY_API_KEYandSTABLY_PROJECT_IDat the job level to avoid repetition - corepack before setup-node - For pnpm and yarn berry (v2+),
corepack enableMUST come beforeactions/setup-nodebecause setup-node needs the package manager binary to set up caching. Yarn classic (v1) does not need corepack. --frozen-lockfile/--immutable- Always use lockfile-strict install in CI to prevent drift- Browser install is separate -
npm cidoes NOT install browser binaries; always add the explicit install step - Monorepo awareness - If playwright config is in a subdirectory, set
working-directoryon the job - Never hardcode secrets - Always use
${{ secrets.* }}references
Capabilities
Install
Quality
deterministic score 0.45 from registry signals: · indexed on github topic:agent-skills · 6 github stars · SKILL.md body (17,527 chars)