## Why
In https://github.com/expo/eas-cli/pull/3406 EAS workflow building the worker was **not** updated to Yarn v4. This breaks building worker.
I tried adding `defaults.tools.yarn: 4.12.0`, but this unfortunately [is **not** the way to use Yarn v4 on EAS as I've learned](https://docs.expo.dev/more/create-expo/#yarn-2-modern). The right way is using Corepack.
However, previously we explicitly did **not** use Corepack, because `pnpm --version` check the worker does in the beginning of its run was failing because it was run without `cwd`, so it was running within worker package which was configured with Yarn.
Since we want to use Yarn v4, we need to figure out a way to use Corepack.
## How
- Enabled Corepack.
- Set up `yarn` executable like the one we have in `universe`.
- Changed `pnpm --version` to **not** run inside worker (so it does not error).
- Removed `setup-node` for specific Node versions in favor of a Mise override.
## Test plan
- Corepack + Yarn resolution checks (local):
- `mise exec [email protected].4 -- bash -lc 'corepack enable; yarn --version'` => `4.12.0`
- same check from `scripts/` => `4.12.0`
- Worker runtime unit test (local):
- `mise exec [email protected].4 -- bash -lc 'corepack enable; yarn workspace @expo/worker test:unit -- runtimeEnvironment.test.ts'`
- GitHub workflow runs:
- success on Corepack worker workflow update: https://github.com/expo/eas-cli/actions/runs/22133707462
- latest run for full set of changes: https://github.com/expo/eas-cli/actions/runs/22133932635
## Why
- Repository maintenance: standardize on Yarn 4 managed via Mise.
- Prepare for introducing minimum package-age constraints in dependency policy.
## How
- Migrated repo tooling to Yarn 4 with Mise.
- Updated CI and scripts for Yarn 4 behavior:
- Lerna invoked via `yarn run -T lerna`.
- Coverage args forwarded correctly (`-- --coverage`).
- Removed unsupported Yarn 4 `--silent` usage.
- Added missing package-local script dependencies surfaced by Yarn 4 (`jest`, `oxlint`, `rimraf`).
- Increased `EasCommand` coverage test timeout from 30s to 60s. (Was consistently failing in CI.)
## Test Plan
- CI only: `Run tests` matrix (including Node 20 coverage) and other touched workflows.
- Release behavior will be verified during the normal release run.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
We're moving to Oxc in universe too. The toolset is faster than ESLint / Prettier.
# How
Asked Codex to migrate the repository. I don't think all rules got migrated, unfortunately, but also the lint are in a bit weird state (different rules in `eas-cli` vs the rest of the repo). I hope we can unify the rules soon.
# Test Plan
CI should pass.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
Our GitHub released executable Oclif bundles of `eas-cli` are… not too popular. v16.28.0 was downloaded 154k times in the last 7 days from NPM. Since its release (2 months ago) it was downloaded from GitHub 82 times.
<img width="1247" height="491" alt="Zrzut ekranu 2026-01-30 o 15 15 35" src="https://github.com/user-attachments/assets/9c5a904a-e00e-43ee-9875-d1cea15b2108" />
# How
Asked Gemini to remove this. This should speed up release process significantly. This release process has been created 5 years ago (https://github.com/expo/eas-cli/pull/54).
Also removed size comparison workflow. It's not going to make comments like https://github.com/expo/eas-cli/pull/3361#issuecomment-3824036144.
Closes https://github.com/expo/eas-cli/pull/3346.
# Test Plan
CI should pass.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
I noticed I didn't migrate all workflows to Mise (because not all worklfows were using Volta lol).
# How
Codex added a shared Mise action (in other repos we have it set up like this too) and used it in all places where we require Node related to code (there's one workflow which does not use checkout, it remains on `setup-node` too). `test.yml` where we test against different Node versions still uses `setup-node`.
Note: this does downgrade Node in a few workflows from 22 to 20, but if 20 is what we decide to use for our development maybe CI should use it too?
# Test Plan
CI should pass.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
I think we migrated from Volta to Mise in… most if not all places?
# How
I think I let Codex do all the work, nothing too difficult. Grepped repo for "volta" leftover afterwards.
One thing I'm not sure of is whether worker is going to work. I remember trying to remove Volta in `turtle-v2` and it wasn't working for some reason? We'll see.
# Test Plan
Compared these changes with https://github.com/expo/universe/commit/acca8169746c5cb4d708add083f77b927341ae4a.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
We moved more packages into this monorepo and the existing test setup didn’t scale well:
- the root Jest config only covered a subset of packages,
- PRs ran heavier “all tests” workflows,
- worker builds ran on PRs even though typecheck + tests were enough.
This PR aligns testing with the dependency graph, reduces unnecessary CI work, and standardizes Jest config handling.
# How
- Switched root testing to Lerna and updated docs.
- CI now runs affected+dependent tests on PRs and all packages on main.
- Coverage runs via Lerna with per-package reports (no merge) and is kept for PRs.
- Worker builds are skipped on PRs; worker workflow triggers include dependency paths on main.
- All Jest configs are now TypeScript with a shared base (`jest/jest.shared.config.ts`) and use `@jest/types` to avoid TS config parsing errors.
- CI stays unit-only (no integration tests).
# Test Plan
Not run locally. CI covers:
- `yarn lerna run test` (affected on PRs, all on main)
- `yarn lerna run test -- --coverage`
`package.sh` currently does not copy workspace packages into the temporary build folder which makes worker pull packages from NPM.
Instead, we want to use local versions from workspace.
Added `copy_package` calls for all dependencies of `@expo/worker`.
It was done in a similar way in `turtle-v2` — https://github.com/expo/turtle-v2/blob/245eaf23492f2dcd9204a4606bb95c109208d29e/src/services/worker/package.sh#L46-L47.
Ran `./package.sh` and confirmed `@expo/eas-build-job` and other packages are symlinked.
## Why
The build-and-deploy-worker workflow had three separate jobs (setup, build-workers, deploy) with complex inter-job dependencies and redundant conditional logic. I think if we put it all in one job it should make stuff faster.
## How
Consolidated all three jobs into a single `build-and-deploy` job with conditional steps:
1. **Check**: Verify if tarballs for the current SHA exist in GCS (when deploying)
2. **Build**: Execute EAS workflow if needed (PR testing or cache miss)
3. **Upload**: Upload tarballs to GCS with SHA-based naming
4. **Deploy**: Copy tarballs to environment-specific names (when deploying)
Key simplifications:
- **Assumption**: Main branch always specifies an environment (staging/production), so `environment == ''` only occurs on PR builds
- **SHA-based caching**: Check cache by SHA regardless of branch - if the code is identical (same SHA), reuse existing tarballs
- **Clearer conditions**: Each step has a single clear purpose without overlapping branch-specific logic
I also added back building Linux tarball on Linux not to mix CPU architectures accidentally.
## Test Plan
[Build on PR](https://github.com/expo/eas-cli/actions/runs/20920297113/job/60104039095?pr=3303). [Deployment from a branch](https://github.com/expo/eas-cli/actions/runs/20921082109/job/60106516235).
Co-authored-by: Claude <[email protected]>
# Why
- Build Android and iOS worker tarballs together in a single EAS workflow run
- Avoid split builds across different machines and simplify uploads
# How
- Update the EAS workflow to build and upload both iOS and Android tarballs
- Update the GitHub Actions workflow to run the EAS workflow once and upload both artifacts
- Cancel the EAS workflow run when the GitHub Actions job is canceled
# Test plan
- Verified auto-cancel behavior: https://github.com/expo/eas-cli/actions/runs/20849238272/job/59900072125
# Why
- Canceling the GitHub Actions workflow should cancel the underlying EAS build/workflow runs to avoid wasting resources.
# How
- Start the iOS build and Android workflow with `--no-wait` and capture their IDs from JSON.
- Poll the EAS build/workflow statuses to determine success/failure.
- Persist IDs to `GITHUB_ENV` and cancel them in a `if: cancelled()` step.
# Test Plan
https://github.com/expo/eas-cli/actions/runs/20859556010/job/59935355509
This PR adds system tests that run after staging deployment and gate production deployment.
## Changes
- Add `system-tests.yml` workflow (can be dispatched manually or called from worker.yml)
- Add EAS workflow fixture for Android build
- Gate production deployment on system tests passing
- Run iOS build (`eas build`) and Android build (`eas workflow:run`) concurrently
## Testing
Once merged, the workflow can be manually dispatched to test, or it will run automatically on the next push to main that touches worker files.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
Gives us more concurrencies for building worker (which is not a huge problem, but may be nice). Also, IF we botched worker staging, we won't break the whole pipeline (system tests won't let worker deploy to production so worker in production will still work and let us rebuild worker).
# How
Changed app ID and token to expo.dev ones.
# Test Plan
CI ran ok.
# Why
The worker integration tests were slow and required platform-specific CI runners (macOS for iOS, Linux for Android) because they attempted to build real React Native apps. This made the CI pipeline complex, slow, and resource-intensive.
# How
- **Replace full-build tests with fast mocked tests**: Instead of running actual Android/iOS builds, we now mock `@expo/build-tools` at the library boundary. This lets us test the worker's WebSocket communication, job dispatch, state sync, abort handling, and hanging worker detection without the overhead of real builds.
- **Add new `jobExecution.test.ts`**: A new integration test that verifies the worker accepts jobs, executes them (via mocked builders), and returns success/error responses correctly.
- **Update existing tests**: `abortBuild.test.ts`, `stateSync.test.ts`, and `possiblyHangingWorker.test.ts` now mock `@expo/build-tools` instead of the worker's internal `build.ts` module, allowing more of the real code path to execute.
- **Simplify CI**: Consolidated the separate `worker-android-checks` and `worker-ios-checks` jobs into a single `worker-checks` job running on `ubuntu-latest`. Removed dependencies on `direnv`, `setup-gcloud`, and platform-specific scripts.
- **Cleanup**: Deleted unused files including `__mocks__/build.ts`, macOS-specific Jest config, CI shell scripts, and the `direnv` internal action.
The build-build part is going to be tested in system tests running in CI on `main`.
# Test Plan
cd packages/worker
yarn test:integrationAll 19 tests pass in ~6 seconds (down from minutes with real builds).
```bash
cd packages/worker
yarn test:integration
```
All 19 tests pass in ~6 seconds (down from minutes with real builds).
## Why
The `worker.yml` GitHub Actions workflow was recently moved from a private repository to this public one. This means we lost access to private GitHub Actions runners and now have to wait in long queues for public `macos-latest` runners when building the iOS worker tarball.
## How
Instead of building the iOS worker tarball on GitHub Actions `macos-latest`, we now:
1. **Trigger an EAS Workflow** from GitHub Actions that builds the tarball on EAS's macOS runners
2. **Upload the tarball as an EAS artifact** using `eas/upload_artifact`
3. **Uploads the artifact to tarball bucket** from GitHub Actions
### Changes
- Added `packages/worker/app.json` - minimal Expo app config required for EAS
- Added `packages/worker/eas.json` - EAS configuration
- Added `packages/worker/.eas/workflows/build-worker.yml` - EAS Workflow that builds the iOS worker tarball and uploads it as an artifact
- Updated `.github/workflows/build-and-deploy-worker.yml`:
- Split the matrix build into separate `build-android` and `build-ios` jobs
- `build-ios` now runs on `ubuntu-latest` and triggers the EAS Workflow using `eas workflow:run --ref --wait --json`
- Extracts the artifact download URL from the workflow output and streams it directly to GCS
## Test Plan
https://github.com/expo/eas-cli/actions/runs/20796579975 Don't mind the CI failure on the PR, I dispatched the workflow manually later.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
We want to streamline DX of the worker part of EAS. Part of it is colocating all code in one repository.
# How
Copied code over to `eas-cli`. Adjusted code — each commit represents a single piece of work towards adjusting code to this repository. The main changes are around removing dependency on `turtle-v2` internal libraries.
One thing to figure out still is deployment process. I've added `build-and-deploy-worker` which is capable of deploying to staging and production. The question is how to automate it. In `turtle-v2` we have system tests. Should we just copy them over to here? (Don't like that.) Should we streamline system tests now and instead of having this weird `turtle-v2-example` repo, create project on the go? Should we deploy worker manually for now?
I think I'd rather deploy manually for now?
# Test Plan
Deployed worker to staging. Ran system tests. They succeeded!! https://github.com/expo/turtle-v2/actions/runs/20042956545/job/57535011098
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
Node 24 is in LTS, it's time to officially support it and test for it.
# How
Add 24 to CI test matrix.
# Test Plan
CI
# Why
Provide a way to move the "latest-eas-build" version tag added in the EAS CLI release process. After https://github.com/expo/universe/pull/22530, this is the EAS CLI version that will be used by EAS workflows.
# Test Plan
Test with different version inputs, ensure failures are handled correctly.
# Why
If a user executes `eas workflow:run --wait`, it would be desirable to show status updates for the individual jobs and steps in the workflow while the workflow run is in progress.
# How
Added code that leverages some of the same utilities used by `workflow:logs`.
During the run, the logs for the various workflow jobs are retrieved and parsed, and the progress of each job is shown in the text of a spinner, showing the current step and the last few lines of the step's logs.
If the workflow run fails, the output will show jobs that failed, along with the failed step for each, and any logs for the failed steps.
Other changes:
- Bumped the "ora" package to the latest version, to pick up bug fixes
- Removed deprecated Node 18 from test runs
- Refactored the code in `workflow:run` that parses inputs, moving that code to a file in `commandUtils`
- Modified the Zod type for inputs to align with server side changes in https://github.com/expo/universe/pull/21950
# Test Plan
- Testing with different workflows
- CI should pass
# Why
The existing release trigger workflow does not obtain the GItHub history and tags, so that Lerna cannot correctly determine which package versions to bump.
# How
Set `fetch-depth: 0` in the checkout action, and do an explicit fetch of all tags.
# Test Plan
- After merge, a dry run of the release trigger should correctly bump only changed packages.
# Why
The new release GitHub workflow has an error because the release script needs to be run without prompts:
```
Checking if eas-cli-local-build-plugin upgrade is available
eas-cli-local-build-plugin is already on the latest version - 1.0.171
lerna notice cli v8.2.3
lerna info ci enabled
lerna info current version 16.17.3
lerna info Assuming all packages changed
Changes:
- eas-build-cache-provider: 16.4.2 => 16.17.4
- eas-cli: 16.17.3 => 16.17.4
- @expo/eas-json: 16.13.2 => 16.17.4
error Command failed with exit code 130.
? Are you sure you want to create these versions? (ynH) info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.
Error: Process completed with exit code 130.
```
# How
- Added the `--yes` flag to the script commands
- Instead of failing when run locally, the script now executes the release flow in "dry run" mode, without committing, tagging, or pushing changes.
- Add workflow input for dry run mode (true by default, must be manually set to false to actually release)
- Get full Git history before running release script, as required for correct Lerna version check
# Test Plan
- Test locally
- Run the workflow in dry run mode after merging
- Attempt a release after this PR is merged
# Why
The current release script requires that the user executing the script have permission to push changes to main without any PR requirement or checks.
# How
To allow us to add PR requirements and checks, I have added a manual GitHub workflow so that the existing script is executed by the expo-bot user, which is exempted from PR requirements.
I have also made a few adjustments to the `release.yml` workflow (remove incorrect release name, update Node version to 22).
# Test Plan
- CI should pass
- Try a release after this PR is merged
# Why
Now that #3091 is resolved, we need to add some CI to guard against similar future issues with GraphQL breakage.
# How
Add a new workflow that
- Executes GraphQL generation, and fails if there is an error.
- Verifies that for a given PR, any changes in GraphQL generated code or schema files have been checked in as part of the PR. If not, the workflow fails.
The new workflow will execute on pushes, and on a daily schedule to guard against breakage due to changes on the server side.
# Test Plan
- CI should pass
- Push a source change in the GraphQL code, but do not regenerate GraphQL, and verify that the new workflow fails
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
When we `git clone` the source repository and then copy worktree changes over, we add and modify files. However, we do not _delete_ files that have been deleted in worktree.
# How
When we operate in `requireCommit: false` mode, we add `--no-checkout` to the `git clone` command. This way the destination directory only has `.git` directory when we start copying files over. Thus, files deleted from worktree will not suddenly appear in tarball.
In `requireCommit: true` mode we don't add `--no-checkout`, because we don't copy files.
# Test Plan
Added test. Tested manually.
Side effect of this is that `git status` reports that all files have been deleted and there's a bunch of untracked files (which, `git add`-ed, make the repository appear as the source repo). I don't know how to copy the Git index 1-1…
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
If you had an `.easignore`, `isFileIgnoredAsync` was always thinking all files are NOT ignored which was causing it to think all builds are in `.GENERIC` mode (with native code) which was not true, because ACTUALLY the files WERE ignored.
All mostly because of forgotten `ignore.initIgnoreAsync()`…
# How
Added `ignore.initIgnoreAsync()`. Also realized whether the file is tracked / not-tracked may affect how actually is it going to behave, so added it to the mix.
# Test Plan
Added tests that I think test all cases.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
It seems like this file includes redundant entries + is not up to date
# How
Remove redundant entries, add @radoslawkrzemien to global codeowners, make @wschurman and @quinlanj codeowners of everything EAS Update related (including channel commands), remove @byCedric and @EvanBacon from EAS Update codeowners
# Test Plan
Review
* chore: update node versions to test with to 18, 20, 22
* update CHANGELOG.md
* chore(eas-cli)!: bump minimum supported Node version to 18
This makes the PR a breaking change.
* docs(eas-cli): add breaking change(log) entry
* chore: update `engines.node` to 18 everywhere
* [ci] bump `@expo/eas-json` version in universe after the release
* Update .github/workflows/release.yml
* test
* edit eas.json
* Revert "edit eas.json"
This reverts commit 3979a1b1fb.
* Revert "test"
This reverts commit 69db39f548.
<!-- If this PR requires a changelog entry, add it by commenting the PR with the command `/changelog-entry [breaking-change|new-feature|bug-fix|chore] [message]`. -->
<!-- You can skip the changelog check by labeling the PR with "no changelog". -->
# Why
It is currently not possible to use `if: ${ always() }` in Custom Build workflows due to outdated `@expo/steps` dependency.
# How
Added a GitHub workflow, similar to the one we have in `turtle-v2`, which will automatically upgrade all dependencies from the `eas-build` repo.
# Test Plan
https://github.com/expo/eas-cli/pull/2220
* [ci] add update eas cli used on production for builds triggered with GitHub trigger once new version is published
* remove unnecessary docs from releasing