For the complete documentation index, see llms.txt. A Markdown version of this page is available by requesting this URL with the header "Accept: text/markdown".
Software Delivery Agent
Agent Overview
Deployments
Builds
InfrastructureDatabasesArtifactsRepositoriesCode Review

Fast and Reliable Testing

Run fewer tests. Trust every result.

Harness combines Test Intelligence, automatic flaky test quarantine, and enforced code coverage — so your pipeline runs faster without losing signal or trust.

Start for FreeRead the docs
80%reduction in tests runavg. across customers
4×faster build timeswith full optimization
0false positivesML tracks all test deps
5+supported languagesJava, Python, Ruby, .NET, Go
Three pillars of reliable testing

Every safeguard runs automatically on every build

Most test suites make you choose between speed and trust. Test Intelligence, flaky test quarantine, and coverage enforcement run together on every build, so the gains compound instead of trading off.

Test Intelligence

AI analyzes your codebase and selects only the tests impacted by a code change.

Flaky Test Detection and Quarantine

Automatically detects and quarantines non-deterministic tests.

Code Coverage

Set a coverage threshold. If a PR drops below it, the pipeline blocks.

Test Intelligence

Only run the tests your change actually touches

Test Intelligence maps every test to the code it exercises, so when a PR lands, your pipeline runs the tests that matter and skips the rest — safely.

Intelligent test selection. Harness analyzes each code change and automatically runs only the unit tests that exercise it, skipping the rest without sacrificing signal.

Multi-language support. Works out of the box with Java, Python, Ruby, C#, Scala, Kotlin, and JavaScript (beta), across frameworks like JUnit, pytest, RSpec, and Maven.

Parallel test execution. Splits the remaining tests into parallel workloads that run simultaneously, compounding the time saved from smart selection.

See Test Intelligence docs

Language and framework support

Test Intelligence works with the languages and frameworks your team already uses.

Java

JUnit 4 & 5, TestNG

Python

pytest, unittest

Ruby

RSpec, Minitest

.NET / C#

NUnit, MSTest, xUnit

Go

Go test

JavaScript

Jest, Mocha (coming soon)

Kotlin

JUnit via JVM

Scala

ScalaTest via JVM

Flaky Test Detection and Quarantine

Stop chasing tests that lie

Tests that pass and fail inconsistently erode trust in your pipeline and waste hours on reruns. Harness automatically identifies flaky tests and quarantines them, so a flaky failure never blocks a good deploy.

Automatic detection

Harness tracks pass/fail history for every test across builds and flags tests with inconsistent outcomes — no manual tagging required.

Quarantine, not deletion

Flaky tests are moved out of the main pipeline gate automatically. They keep running and reporting, but they can no longer block a merge.

Root cause visibility

See flake rate, failure history, and last known-good run for every quarantined test, so your team can fix the underlying issue instead of ignoring it.

See test management docs

Flaky tests · Last 30 days

AuthServiceTest.testConcurrentLogin38%
19 of 50 runsQuarantined
SearchIndexTest.testRebuild27%
13 of 48 runsFlagged
CacheTest.testExpiry15%
7 of 46 runsMonitoring
3 tests quarantined this month — 0 blocked merges from flaky failures
Code Coverage

Know exactly what your tests protect

Code Coverage runs on every build, down to the file, so you always know what lines are covered by tests and what's not.

Visibility.

See line and branch coverage on every pull request, down to the file level, so gaps surface before they ship.

Checks.

Set a configurable coverage threshold as a pipeline gate. PRs that drop below it fail the status check before they merge.

Insights.

Track coverage trends by repo and branch over time even when Test Intelligence runs a subset of tests.

See code coverage docs
Get started

Cut your test suite time by 80%

Test Intelligence takes minutes to configure and starts delivering results on the first build. No test refactoring required.

Start for Free Read the docs