Start free
Andrew Hanna

Andrew Hanna

sfdx-hardis and the Open-Source Salesforce DevOps Stack: An Honest Read

sfdx-hardis and the Open-Source Salesforce DevOps Stack: An Honest Read

The open-source Salesforce DevOps stack is real, mature and free, and it covers more than most people assume. sfdx-hardis wraps the Salesforce CLI into an actual pipeline, sfdx-git-delta computes deltas, Code Analyzer handles static analysis, and your CI platform runs it. What it does not give you is an owner, a system of record, or someone to call at 2am. That is the half teams end up building themselves.

What is the open-source Salesforce DevOps stack?

Roughly five layers, each independently maintained:

  • Orchestration. sfdx-hardis, a free and open-source layer over the Salesforce CLI, led by Nicolas Vuillamy and Cloudity. It defines CI/CD pipelines, daily metadata backup and org monitoring, and AI-generated project documentation. It runs on GitHub, GitLab, Azure DevOps and Bitbucket, and installs as an sf plugin, a VS Code extension, or Docker images.
  • Delta computation. sfdx-git-delta, the community plugin that turns a Git diff into a package.xml and a destructiveChanges.xml.
  • Quality gates. Salesforce Code Analyzer, open source and CLI-native, unifying PMD, ESLint, CPD, RetireJS, a regex engine and a Flow scanner behind one YAML config.
  • Modular builds. sfp from flxbl, the successor to DxAtScale's sfpowerscripts, for teams that model their org as packages.
  • The runner. GitHub Actions, GitLab CI, Azure Pipelines or Jenkins, doing the actual executing.

Assembled well, that is a serious pipeline. Anyone claiming you cannot do Salesforce DevOps without a licence has not looked recently.

What does sfdx-hardis actually cover?

More than a CLI wrapper. Three things stand out.

First, it takes the admin problem seriously. Its VS Code extension lets a non-developer produce a pull request from clicks, which is a deliberate answer to the fact that most Salesforce teams are not developers.

Second, it treats monitoring as a first-class job, not a side effect: scheduled metadata backup and org monitoring run in their own repository, separate from delivery.

Third, it moved early on AI agents. Over 130 of its commands accept an --agent flag for non-interactive execution by coding agents, which is a genuinely forward-looking design decision for a free tool.

What does open source cover well?

  • Deploying metadata between orgs, with delta and destructive changes handled properly.
  • Static analysis and quality gates in CI.
  • Scheduled backups and drift monitoring.
  • Package builds for teams that already think in packages.
  • Cost. There is no licence, and for a team with one motivated engineer that matters.

If your bottleneck is any of those five, the honest answer is that you probably do not need to buy anything.

Where do teams end up building the missing half themselves?

This is where the conversation usually gets dishonest in both directions. So, specifically:

An owner

Every layer above is a dependency you maintain. Node upgrades, CLI breaking changes, plugin versions, a Docker image that has drifted. It is not much work in any given month, and it is always the same person. When that person leaves, the pipeline becomes a haunted house.

A system of record

CI logs tell you what a job did. They do not tell you which change is currently in UAT, when it got there, who approved it, or what shipped in the last release. Teams rebuild this in a spreadsheet, a Jira board, or a small internal app. It is the single most commonly rebuilt piece.

Access control that is not the CI platform's

Who may deploy to production, who approves, who sees which project. Branch protection and environment rules cover part of it. Per-project permissions, SSO and an audit log that a compliance reviewer accepts are a different piece of work.

A front door for non-developers

sfdx-hardis addresses this better than most, and a VS Code extension is still an install, an update path and a mental model. Teams whose majority is admins and consultants often end up wrapping the pipeline in something ticket-shaped anyway.

ISV and package delivery beyond the build

Building a 2GP package is solved. Knowing which version each subscriber org runs, resolving cross-package dependencies, and gating an AppExchange release are not the same problem, and the open-source stack largely leaves them to you.

Someone to call

Open source gives you a community and source code, which is more than a vendor gives you in some ways and less in one: there is no SLA on your release night.

Who should run the open-source stack, and who should not?

Run it if you have an engineer who wants to own delivery tooling, your team is comfortable with Git and YAML, and your org-to-org flow is the whole job. That describes a lot of good teams, and they should not feel bad about it.

Buy instead if most of your team never opens a terminal, if nobody wants to own the pipeline as a permanent side job, or if you need audited access control and subscriber-level package tracking. Not because the open-source tools are weak, but because you would be building the missing half with people whose day job is something else.

That is the gap Serpent is built for: ticket-based deployments with Git in the background, AI code review on every plan, native 1GP, 2GP and AppExchange release workflows, version management across subscriber orgs, and a native MCP server so agents drive deployments through preflight checks and human approval. Flat per-company pricing, and a free tier if you want to test the claim.

Open source is not the enemy here. It is the reference implementation, and it keeps the whole category honest about what should be table stakes.

FAQ

Is sfdx-hardis free?

Yes. It is open source with no licence cost, maintained by Cloudity and community contributors, and Cloudity offers paid professional services around it.

Can sfdx-hardis replace Copado or Gearset?

For org-to-org delivery with a technical owner, it covers the same ground. The difference shows up in governance, non-developer access and vendor support rather than in deployment mechanics.

Do I need the Salesforce CLI to use it?

Yes. It runs on Node.js and the Salesforce CLI, and installs as an sf plugin, a VS Code extension or a Docker image.

Is Salesforce DevOps Center open source?

No. It is free and Salesforce-native, which is a different thing. Open source means you can read and change the code.

What is the real cost of the free stack?

Maintenance time and key-person risk. Budget a named owner and a few hours a month, and it is a fair trade. Leave it unowned and it stops being free.

Related Articles

Curious about faster shipping before you dive in? Let's talk

Commitment free!