A custom Vitest reporter that generates Flakiness Reports from your Vitest test runs. The reporter automatically converts Vitest test results into the standardized Flakiness JSON format, capturing test outcomes, system utilization, and environment information.
- Requirements
- Installation
- Quick Start
- Uploading Reports
- Viewing Reports
- Features
- Configuration Options
- Environment Variables
- Example Configuration
- Node.js 20.17.0 or higher (22.9.0 or higher on Node 22)
- Vitest 4.0 or higher
- Node.js project with a git repository (for commit information)
- Valid Flakiness.io access token (for uploads)
npm install @flakiness/vitestAdd the reporter to your vitest.config.ts:
import { defineConfig } from 'vitest/config';
export default defineConfig({
test: {
includeTaskLocation: true,
reporters: [
'default',
['@flakiness/vitest', { flakinessProject: 'my-org/my-project' }]
],
},
});Note
The flakiness reporter should be added alongside your other reporters. Include 'default' (or another built-in reporter) to retain the standard terminal output, since Vitest only uses the reporters listed in the reporters array.
Tip
Setting includeTaskLocation: true is recommended to enable test locations in the report.
Run your tests. The report will be automatically generated in the ./flakiness-report folder:
npx vitest runView the interactive report:
npx flakiness show ./flakiness-reportReports are automatically uploaded to Flakiness.io after test completion. Authentication is resolved in the following order, and the first available method wins:
-
Access token: Provide a token via the
tokenoption or theFLAKINESS_ACCESS_TOKENenvironment variable. -
GitHub OIDC: When running in GitHub Actions with no access token, the reporter can authenticate using GitHub's OIDC token. This requires three conditions:
- The
flakinessProjectoption must be set to your Flakiness.io project identifier (org/project). - The Flakiness.io project must be bound to the GitHub repository that runs the GitHub Actions workflow.
- The workflow must grant the
id-token: writepermission.
- The
-
GitLab OIDC: When running in GitLab CI/CD with no access token, the reporter can authenticate using a GitLab ID token. GitLab mints ID tokens when the job starts, so the job must declare one named
FLAKINESS_ID_TOKENwhose audience is your project identifier:test: id_tokens: FLAKINESS_ID_TOKEN: aud: my-org/my-project # must match the `flakinessProject` option script: - npx vitest run
The
flakinessProjectoption must be set, and the Flakiness.io project must be bound to the GitLab project running the pipeline.
If no method is available the upload is skipped, and if an upload fails the report is still available locally in the output folder. Either way the test run is unaffected.
After test execution, you can view the report using:
npx flakiness show ./flakiness-reportSee features.md for this reporter's status against the Flakiness Report spec.
When includeTaskLocation: true is set in your Vitest config, the reporter records the exact file, line, and column for each test. This enables precise navigation from the Flakiness.io dashboard back to your source code.
Browser Mode records screenshots as Vitest artifacts. The reporter writes them into the local Flakiness report and uploads them along with it. Identical content is stored only once, no matter how many tests attach it.
Two kinds of screenshot are captured:
- Failure screenshots. Vitest screenshots every failing test when
browser.screenshotFailuresis enabled — which is the default outside of the Vitest UI. It is reported asfailure-screenshot.png, and needs no changes to your tests. - Visual regressions. Each failed
toMatchScreenshot()comparison contributesscreenshot-1-expected.png,screenshot-1-actual.pngandscreenshot-1-diff.png, numbered per comparison within the test. Vitest does not take a failure screenshot for these, since the comparison images already show the page.
Attachments of custom artifacts recorded through Vitest's recordArtifact() are named after the artifact type (my-plugin:artifact-1). Attachments of test annotations (context.annotate()) are not collected.
Vitest allows creating tests with identical names:
// Trivial duplicates
it('should work', () => { /* ... */ });
it('should work', () => { /* ... */ });
// More common way to end up with test duplicates:
it.for([
{ input: 1, expected: 2 },
{ input: 1, expected: 3 },
])('should handle $input', ({ input, expected }) => { /* ... */ });Flakiness.io, however, does not allow test duplicates.
Flakiness.io considers two tests to be duplicates when they:
- Share the same parent suite hierarchy
- Have the same test name
- And run in the same Vitest project
Flakiness.io relies on full test names to construct test history, so each test must have a unique full name. When Vitest reporter detects tests with identical full names, it issues warnings and handles them according to the duplicates option.
For each Vitest project, the reporter creates a unique environment. If a project has a name, it will be used as the environment name in Flakiness.io. Tests run in different projects get separate history timelines.
Environment variables prefixed with FK_ENV_ are automatically included in the environment metadata. The prefix is stripped and the key is converted to lowercase.
Example:
export FK_ENV_DEPLOYMENT=staging
export FK_ENV_REGION=us-east-1This will result in the environment containing:
{
"metadata": {
"deployment": "staging",
"region": "us-east-1"
}
}Flakiness.io will create a dedicated history for tests executed in each unique environment. This means tests run with FK_ENV_DEPLOYMENT=staging will have a separate timeline from tests run with FK_ENV_DEPLOYMENT=production, allowing you to track flakiness patterns specific to each deployment environment.
The reporter automatically detects CI environments and includes:
- CI run URLs (GitHub Actions, Azure DevOps, Jenkins, GitLab CI)
- Git commit information
- System environment data
The reporter accepts the following options:
Optional human-readable report title. Typically used to name a CI run, matrix shard, or other execution group. Defaults to the FLAKINESS_TITLE environment variable, or empty otherwise.
reporters: [
['@flakiness/vitest', { title: 'Shard 1/4 — Linux Chrome' }]
]The Flakiness.io project identifier in org/project format. Required for CI OIDC authentication. When set, and when the Flakiness.io project is bound to the repository running the pipeline, the reporter authenticates uploads via a GitHub Actions or GitLab CI/CD OIDC token with no access token required. See Uploading Reports for the per-provider requirements.
reporters: [
['@flakiness/vitest', { flakinessProject: 'my-org/my-project' }]
]Custom Flakiness.io endpoint URL for uploading reports. Defaults to the FLAKINESS_ENDPOINT environment variable, or https://flakiness.io if not set.
Use this option to point to a custom or self-hosted Flakiness.io instance.
reporters: [
['@flakiness/vitest', { endpoint: 'https://custom.flakiness.io' }]
]Access token for authenticating with Flakiness.io when uploading reports. Defaults to the FLAKINESS_ACCESS_TOKEN environment variable.
If no token is provided, the reporter falls back to CI OIDC on GitHub Actions and GitLab CI/CD. If that is unavailable too, the report is still generated locally but won't be uploaded automatically.
reporters: [
['@flakiness/vitest', { token: 'your-access-token' }]
]Directory path where the Flakiness report will be written. Defaults to flakiness-report in the current working directory, or the FLAKINESS_OUTPUT_DIR environment variable if set.
reporters: [
['@flakiness/vitest', { outputFolder: './test-results/flakiness' }]
]Controls how the reporter handles tests with duplicate full names. Defaults to 'fail'.
Warning
The 'rename' mode is not recommended for regular use. There is no guarantee that test histories will remain stable for duplicate tests, since the renaming is based on internal Vitest identifiers that may change between runs. This mode exists to help evaluate the reporter against large Vitest projects that have not yet resolved their duplicate test names.
'fail'(default): Duplicate tests are marked as failed with a descriptive error message and adupeannotation. The first duplicate gets a single failed attempt explaining the problem; the remaining duplicates are stripped of their attempts (effectively hidden from the report). This is the recommended mode — it surfaces the problem so you can fix it by renaming your tests.'rename': Duplicate tests are automatically renamed by appending a suffix (e.g.,– dupe #2,– dupe #3) to their titles. Each renamed test also receives adupeannotation on all its attempts. The first test with a given name keeps its original title; only the subsequent duplicates are renamed.
reporters: [
['@flakiness/vitest', { duplicates: 'rename' }]
]When set to true, prevents uploading the report to Flakiness.io. The report is still generated locally. Can also be controlled via the FLAKINESS_DISABLE_UPLOAD environment variable.
reporters: [
['@flakiness/vitest', { disableUpload: true }]
]The reporter respects the following environment variables:
FLAKINESS_TITLE: Human-readable report title (equivalent totitleoption)FLAKINESS_ACCESS_TOKEN: Access token for Flakiness.io uploads (equivalent totokenoption)FLAKINESS_ENDPOINT: Custom Flakiness.io endpoint URL (equivalent toendpointoption)FLAKINESS_OUTPUT_DIR: Output directory for reports (equivalent tooutputFolderoption)FLAKINESS_DISABLE_UPLOAD: When set, disables report uploads (equivalent todisableUploadoption)FLAKINESS_ID_TOKEN: GitLab CI/CD ID token used for OIDC authentication. Set by GitLab for jobs that declare it underid_tokens:(see Uploading Reports)
Here's a complete example with all options:
import { defineConfig } from 'vitest/config';
export default defineConfig({
test: {
includeTaskLocation: true,
reporters: [
'default',
['@flakiness/vitest', {
title: 'My Test Run',
flakinessProject: 'my-org/my-project',
endpoint: process.env.FLAKINESS_ENDPOINT,
token: process.env.FLAKINESS_ACCESS_TOKEN,
outputFolder: './flakiness-report',
duplicates: 'fail',
disableUpload: false,
}]
],
},
});