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".

Harness Cloud Runners

手間をかけずに、超高速かつセキュアな CI ビルドを実現

分単位の課金で高速なクラウドビルドを実現。インフラ管理は不要で、予算を圧迫せずにエンタープライズ級のスピードとセキュリティを両立します。

76%infrastructure cost reductionwith Harness Cloud
80%reduction in tests runon average, with Test Intelligence
4×faster incremental buildsvs. cold builds, with Speed Intelligence
20%of engineering time reclaimedno self-hosted build agents to maintain
Harness Cloud Runners

AI-native cloud builds

Harness agents generate pipelines, pick the right tests, and root-cause failures, all on fully managed Harness Cloud Runners. You build software while Harness handles the infrastructure.

Agent-generated pipelines. Harness agents generate pipelines and root-cause build failures so your team spends less time on YAML and logs.

Smarter test runs. Test Intelligence runs only the tests your change affects, so every build minute counts.

Pay for what you use. Simple per-minute pricing with no reserved capacity and no idle waste. Pre-purchase credits for volume discounts.

Hybrid flexibility. Mix Cloud Runners with your self-hosted infrastructure, and keep sensitive builds on-prem from the same pipeline.

オートスケーリング

Linux

x86_64 & ARM

Windows

.NET & PowerShell

macOS M4

Xcode native

並列ステージ・オートスケーリング
Scale

Right-sized machines that scale with you

Cloud-optimized machines scale with your workload, from lightweight unit tests to high-memory Docker builds, so pipelines run smoothly regardless of volume.

Autoscaling. Capacity grows and shrinks with demand, with concurrent stages, matrix, and looping strategies for complex pipelines.

Multi-OS and architecture. Build on Linux, Windows, ARM, and macOS M4 from the same pipeline definition, with no separate farms to maintain.

Resource classes. Match compute to the workload, from lightweight machines for unit tests to high-memory machines for integration tests.

Resource visibility. CPU and memory metrics in the execution view help you right-size machines without leaving your pipeline.

Speed

Fast by default

Isolated machines, pre-cached toolchains, and Test Intelligence keep builds fast without tuning.

Isolated machines. Every build gets its own machine, so there are no noisy neighbors, no cross-build contamination and no performance variance.

Smart caching. Dependencies, Docker layers, and toolchains for popular languages and frameworks are cached automatically.

Test Intelligence. Reduce test time by up to 80% by running only the tests that matter.

High availability. Industry-leading build performance backed by availability SLAs.

スピード最適化
Test Intelligenceテスト実行の高速化
80%
スマートキャッシュ依存関係の再利用
3x
分離されたマシン他ジョブからの影響
0
分離されたランタイム
Build A新規 VM
CPU
メモリ
ディスク
Build B新規 VM
CPU
メモリ
ディスク
Build C新規 VM
CPU
メモリ
ディスク

各ビルドには専用マシンが割り当てられ、使用後は破棄されます。

Security

Secure & isolated cloud services

Every build runs on a fresh, dedicated machine that is destroyed after use, so nothing is shared between builds or customers.

Fresh isolated runtimes. A new machine for every build, destroyed after use.

No noisy neighbors. Dedicated resources mean consistent build performance.

Private, encrypted access. Code travels over SSH and HTTPS, and builds reach on-premises resources through Tailscale (recommended), VPN, or PrivateLink.

Egress controls. Restrict which sites, URLs, and paths your builds are allowed to access.

Harness Cloud Runners vs. self-hosted agents

The hidden cost of self-hosted build infrastructure is the 20% of engineering time spent maintaining it.

FeatureHarness Cloud RunnersSelf-Hosted Agents
Infrastructure
Infrastructure management
Zero setup, fully managed
Patch and maintain agents manually
Autoscales to demand
Instant, unlimited concurrency
Manual scaling required
Fresh machine per build
Isolated VM, no shared state
Shared-state contamination risk
Mac, Linux & Windows
All three in one pipeline
Separate farm per OS
Performance & Reliability
Consistent build performance
No noisy neighbors
Shared pools cause variance
SLSA Level 3 compliance
With Harness SCS, no extra tooling
Manual setup required
Cross-tenant security isolation
Dedicated VM per build
Shared agent pool vulnerabilities
Team Productivity
Infrastructure overhead
0% of engineering time
~20% of engineering time
Time to first build
Minutes to set up
Days to weeks of setup
Full support
Partial support
Not available

Harness Cloud Runners を始める

数分で最初のビルドを実行。インフラの構築や管理は不要です。