> ## Documentation Index
> Fetch the complete documentation index at: https://momentic.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Quarantine
> Skip flaky tests without failing pipelines, manually or via automated rules.
Quarantine is a state that tells Momentic: "keep running this test, but don't
let failures fail the build." It's the right move for tests that are genuinely
flaky or mid-investigation, better than skipping entirely because you still get
data.
When a test is quarantined:
* It still runs on every `momentic run` and every CI job
* Failures are reported as **quarantined failures**, not regular failures
* The CLI returns a non-failing exit code even if quarantined tests fail
* Runs still show up in the dashboard so you can investigate
## Quarantine a test
### CLI
```bash npm theme={null}
npx momentic quarantine add --reason
```
```bash yarn theme={null}
yarn dlx momentic quarantine add --reason
```
```bash pnpm theme={null}
pnpm dlx momentic quarantine add --reason
```
See [`momentic quarantine`](/cli-reference/momentic/commands/quarantine) for the
full command reference.
### Dashboard
Go to **Quarantine** in the sidebar and click **Quarantine test**. You can also
quarantine and unquarantine directly from the run viewer, or hover a test on the
Quarantine page and pick **Unquarantine**.
## Quarantine rules
Rules automate quarantine/unquarantine based on thresholds so you don't have to
intervene manually.
Each rule has:
* **Action**, quarantine or unquarantine
* **Metric**, one of:
* **Pass rate**, percentage of runs that passed
* **Failure count**, absolute number of failed runs
* **Flake rate**, percentage of runs that flaked
* **Recovery count**, absolute number of runs that passed via
[failure recovery](/reliability/failure-recovery)
* **Recovery rate**, percentage of runs that passed via failure recovery
* **Threshold**, value that triggers the action
* **Window**, period over which the metric is evaluated
### Filters
Scope a rule to specific tests or runs:
* Repository
* Branch (useful for quarantining tests failing on `main`)
* Labels (useful for stricter rules on a subset of tests)
