CI, CD and infrastructure

Used here, in this category

• GitHub Actionschecks
  Every gate runs with if: !cancelled(), because a job that stops at the first failure was hiding the checks behind it.
  Evidence: .github/workflows/
• Wranglerserves
  Called directly rather than through an action wrapper: one fewer third party in the release path.
  Evidence: .github/workflows, pinned rather than floating

All 34

Alphabetical. Checked on 2026-09-04; anything that did not answer says so.

• Abstra
  Automation for finance operations, built as workflows rather than as spreadsheets.
  Aimed at the spreadsheet that became a system.
• Ansible
  Agentless configuration management over SSH, with playbooks in YAML and no daemon on the target.
  Won by requiring nothing on the machine it configures.
• AWS Step Functions
  Orchestrates AWS services as a state machine, so retries and failure branches are configuration rather than code.
  The retry logic you keep rewriting, declared once.
• Buildstash
  Artifact and release management for teams shipping binaries: mobile, desktop, games, XR and embedded.
  Built for the builds a container registry was never meant to hold.
• Camunda
  Process orchestration for long-running work that spans people, systems and now agents, with the model as the executable.
  BPMN, still standing, now aimed at agents.
• CircleCI
  A hosted CI platform, one of the older ones, now marketing itself around validating machine-written changes.
  Validate everything, assume nothing.
• Datadog
  Metrics, logs, traces and a great deal more in one platform. The default answer for large-scale monitoring, and priced like it.
  You will find everything, and you will find the bill.
• DeployHQ
  Deploys from a git host to servers with build steps, zero-downtime releases and one-click rollback.
  Deployment for the estate that is not on Kubernetes.
• DeployRamp
  Wraps risky pull requests in feature flags, rolls them out gradually while watching errors, then removes the flag.
  Automates the flag you were going to forget to delete.
• Depot
  Remote build infrastructure with persistent caching, aimed at container and CI builds that repeat most of their work.
  Sells the cache, which is where the time actually goes.
• Docker
  Made the container image a unit of distribution. The reason a build on one machine runs on another.
  Changed what 'it works on my machine' even means.
• GitHub Actions
  CI that lives in the repository it builds. Workflows are YAML in the tree, and the marketplace supplies the steps.
  Won CI by already being where the code was.
  Used here: .github/workflows/
• GitLab
  Source control, CI, registry, security scanning and issue tracking as one application rather than as an assembled toolchain.
  Argues the whole toolchain should be one product.
• GitLab AI the page returns 404
  GitLab's AI features, listed separately. That page now returns 404 and the capability sits inside the main product.
  Folded into the main product.
• Harness
  A delivery platform covering pipelines, feature flags, cost and verification, with automated rollback on bad deploys.
  Tries to own everything between commit and production.
• Infracost
  Shows the cost of an infrastructure change in the pull request, before it is applied.
  Puts the invoice in the diff.
• Jenkins
  The open-source automation server that most of the industry's CI ran on, and much of it still does.
  Older than most of this list, still deployed everywhere.
• Jenkins X
  Opinionated Kubernetes-native CI/CD built on Jenkins. The domain now serves unrelated content.
  The domain moved on.
• Kubernetes
  The container orchestrator that became the industry's default substrate, and the reason platform engineering is a job.
  Won, and created an entire economy of things to make it bearable.
• LambdaTest
  Cloud browser and device grid for running tests across many real environments in parallel.
  Rents the browsers you cannot install.
• Lil'bots
  A playground for small bots and scripts, deliberately informal.
  Hack around and find out.
• n8n
  Workflow automation you can self-host, with code steps where the visual editor runs out, and model nodes throughout.
  The automation tool you can run inside your own network.
• New Relic
  One of the original application performance monitoring platforms, now covering the whole telemetry stack.
  Was measuring response times before observability had a name.
• Ona
  Cloud software engineers running under orchestration and kernel-level isolation, from the team behind Gitpod.
  Isolation at the kernel, because the agent runs code.
• PagerDuty
  Incident response and on-call scheduling. The product that made paging a discipline rather than a phone tree.
  Defined what being on call means.
• Pulumi
  Infrastructure as code in general-purpose languages, so loops, types and tests come from the language rather than from a DSL.
  Infrastructure in a language that already has a test runner.
• Pulumi AI
  Pulumi's agent for provisioning, governing and optimising infrastructure with enterprise controls.
  An agent with permission to change infrastructure, which is the interesting part.
• RunMyJob
  CI runners billed by real usage rather than idle time, compatible with GitLab and GitHub workflows.
  Attacks the idle runner bill.
• Spacelift
  Orchestration for infrastructure pipelines across Terraform, OpenTofu and Ansible, with policy applied between plan and apply.
  The approval gate that Terraform never shipped with.
• Sysdig
  Cloud security from runtime back to source, built on the open-source Falco project for kernel-level detection.
  Detection from the kernel, not from the logs.
• Terraform
  The declarative infrastructure tool that made state files and plan output an industry convention.
  Everyone else is compared to this.
• Terraform Cloud
  Hosted Terraform with remote state, policy and a standard workflow. Now part of the HashiCorp platform.
  State management, which is where self-hosted Terraform hurts.
• Terrateam
  GitOps-native Terraform and OpenTofu automation, driven from pull requests rather than a separate console.
  The plan output belongs in the pull request.
• Trigger.dev
  Open-source background jobs and workflows in TypeScript, with retries, queues and long-running tasks that survive deploys.
  Long-running work that does not die on redeploy.

Other shelves