Changelog

Historically, when something fails on our side, our infrastructure, not yours, our API still responds with HTTP 200. The response body says "internal error," but the status code says everything was fine. Any monitoring tool watching status codes never sees the failure at all.

This is just to let you know we’re fixing that. Internal errors will soon return HTTP 500. Your monitoring catches what it should have caught all along, and so does ours, which means faster detection and faster fixes on infrastructure issues on our end.

Nothing about the failure itself changes, only the status code does. These are errors you could never have fixed on your end anyway: the error message never gave you anything actionable, and that hasn't changed.

We're rolling this out behind a feature flag, and aren’t going to start for about a month, so we can control the pace and revert immediately if needed. This also gives you some warning period about the change. This won’t affect every account in Spacelift, due to configurations, and usage, but there are some things to look for.

What to check

This is technically a breaking change, even though we expect the real-world impact to be small for most teams. It only matters if your integration reads HTTP status codes to drive logic, retries, alerts, pipeline gates, and so on.

• If you have custom automation built on our API, check whether it treats a 200 as "succeeded" without also checking the response body for errors.

• A request that previously came back as 200 during an internal error will now come back as 500.

• Everything else about the request and its outcome is unchanged.

If your integration already handles 5xx responses the way it should, you won't notice this change at all. Have questions? Please let us know. We’re here to help.

That’s right! Now you can ask the AI to deploy a module by name. Spacelift Intent resolves it against your Spacelift module registry and applies it as a single managed resource, under your existing policies, state, and audit trail. No module block, and no PR required. The guardrails already built into your registry and your Intent policies still apply.

"Deploy the elb module" resolves to the module's slug (terraform-aws-elb), picks its latest active version, and prompts for any missing inputs. Pin a version or pass inputs up front for precision: "Deploy the elb module at version 1.2.0, with the web load balancer inputs.

Before applying, Intent verifies the module is visible and active, and that every required input is set with nothing undeclared. If it fails a check then nothing will run. Pass them, and Intent evaluates your Intent policies (resource_type: spacelift/module, operation, provider: the slug, provider_version, proposed_state: your inputs), then runs the module on a worker in the background.

Updates and upgrades work the same way: "Set the idle timeout to 120" or "Upgrade to the latest version." Intent diffs the input schema between versions, and carries your existing values forward. It only asks about what's new or newly required.

Each deployment is one spacelift/module resource in your Intent project. Its managed resources are nested underneath and read-only. Every create, update, and delete action is logged in History.

Intent only deploys modules from your Spacelift module registry. It does not pull modules from the public Terraform Registry or directly from Git. The module has to be published to Spacelift first. As always, take it for a spin, and let us know what you think.

Check out the documentation here: Deploying Spacelift Modules

Spacelift now supports SCIM 2.0 (System for Cross-domain Identity Management). Your identity provider becomes the single source of truth for who has access to Spacelift. Okta, Microsoft Entra ID, OneLogin, or any IdP that speaks SCIM 2.0 can automatically sync users and groups, so you never have to manually invite or remove anyone again.

When someone joins your company or switches teams, their Spacelift access updates automatically based on IdP group membership. When someone leaves, their account is deactivated and all active sessions are revoked. No stale accounts, no forgotten cleanup.

Groups sync too. Push them from your IdP via SCIM, then assign roles using the Spacelift Terraform provider (recommended) or the UI. You get version-controlled, auditable access management that stays current with your IdP automatically.

How it works

Generate SCIM credentials in Organization Settings > Single Sign-On, configure your IdP with the base URL and OAuth 2.0 client credentials, and syncing starts immediately. The credentials are scoped exclusively to SCIM endpoints. SCIM creates user and group records but does not assign permissions. You control what people can do through role bindings or login policies.

Details

• Full CRUD for users and groups via SCIM 2.0 (RFC 7644)

• User deactivation and deletion both revoke active sessions immediately

• Group membership syncs automatically, keeping role assignments current

• Filtering by userName, active status, email (users) and displayName (groups)

• Enterprise plan only. Requires SSO to be configured first. Some IdPs (like Okta) require SAML, not OIDC, for SCIM.

Read the full documentation: SCIM documentation

Have you been working hard since the new learning modules in the Launchpad came out? We know a lot of you have. And now, we’ve introduced a way to commemerate all your hard work, and celebrate your accomplishments! As you complete the learning modules, you can now view and download a certificate of completion. Your whole team can share in the pride as everyone gains more skills by completing missions, and collecting more certificates!

This update also brings the Launchpad and learning modules to even more users of the Spacelift platform. It’s not just admins who can see the Launchpad now. So be sure to check it out, complete some missions, and let us know what you think! Happy Learning!

Please Note: Some missions may not be available to all tiers, and some users may not have all the access needed to complete all of the missions.

That’s right! You can now connect your LLM of choice directly to Spacelift using the new Remote MCP Server. If you already use AI-powered tools or agents for infrastructure work, this gives them a way to take action inside Spacelift, not just talk about it.

The Remote MCP Server works with any LLM or agent that supports the Model Context Protocol. Point it at Spacelift, and your existing tools can create and manage runs, query stack status, trigger deployments, and work with your infrastructure the same way you do in the UI or CLI.

A quick reminder on Infra Assistant modes

If you have not tried Infra Assistant recently, it now has two distinct modes:

• Ask is read-only. You can query your infrastructure, ask questions about stack configurations, check run history, and get answers without changing anything. Good for investigation and context-gathering.

• Build lets Infra Assistant take action. Give it instructions and it will create resources, trigger runs, and make changes in your infrastructure on your behalf. This is where the Remote MCP Server becomes especially useful: your own agents can operate in Build mode through the same interface.

Getting Started

Check out the docs for setup instructions and supported operations. If you run into questions or want help connecting a specific tool, reply to this email and we will get you sorted.

We've added a new and improved Learn section to the Spacelift Launchpad! It's a four-module, hands-on training experience built directly into the platform.

Note: This feature requires admin rights, as most of the tasks would require them. 

What's new

Launchpad Learn includes four modules:

• Foundations — Essential concepts and first steps with Spacelift

• Operational Safety — Policies, approvals, alerts, and audit visibility for safe, compliant changes

• Configuration Reuse — Reusable config, automated setup, and templatized workflows for team-scale consistency

• Delivery at Scale — Multi-stack coordination, dependency management, and confident environment promotion

Each module is divided into sections with sequential tasks. Tasks within a section must be completed in order, but you can move between sections freely. Every step includes code examples and everything else you need to complete it.

Who it's for

Launchpad Learn is designed for users at all levels. New users can work through Foundations to get operational quickly. Experienced users can start with whichever module matches what they're working on.

Where to find it

Go to Launchpad in your Spacelift account and open the Learn section. Check out the video here! 

We're making an improvement to how plan policies are evaluated for Terraform, OpenTofu, and Terragrunt stacks. Currently, when a plan is evaluated against your plan policies, we include all resources in the payload, including resources with no changes (no-ops).

What is changing?

Going forward, we will no longer include unchanged resources in the policy input data. For the vast majority of customers, this change will be invisible. Plan policies that check for planned changes to resources will continue to work exactly as before.

If your plan policies reference or inspect unchanged resources (no-ops), you may need to update your policy logic. We've proactively identified accounts that could be affected and will be reaching out directly.

Why is it changing?

For most stacks, no-op resources make up the large majority of the plan payload. Removing them significantly reduces the size of data processed during policy evaluation, which means faster policy evaluation and better overall performance.

When is it changing?

The gradual rollout will start on the 1th April 2026. Have questions? Feel free to reach out. We’re more than happy to answer a questions, and assist in any way we can.

Today is a big day for us here at Spacelift. We’ve been working hard on some very exciting new functionality. Today, we are pleased to announce Spacelift Intelligence, a set of AI capabilities built directly into the Spacelift platform. It includes two components that work together: an AI Infrastructure Assistant we’re calling Infra Assistant and a natural language deployment model called Intent.

Why This, Why Now?

IaC and GitOps are still the right foundation for managing infrastructure. That's not changing. But those workflows were designed for production. As development velocity increases, especially for experiments, sandboxes, and POCs, the IaC and GitOps pipeline starts to get clogged. AI tools have accelerated the development cycle, but infrastructure hasn’t kept up.

Spacelift Intelligence is our answer to closing this new AI-velocity gap. The goal isn't to replace your Terraform code or your GitOps workflow. It's to give you an AI-native operations layer that gives your platform team the speed of AI without sacrificing the governance, visibility, and control you depend on.

Infra Assistant

Infra Assistant is an AI chat interface embedded in Spacelift. You can ask it infrastructure-related questions and get guidance without leaving the platform.

• “Please find any public s3 buckets across all my stacks”

• “Do I have any CVE vulnerabilities in my environment?”

• “What would you recommend I do to optimize my production deployments?”

Infra Assistant is context-aware, understands your stacks, runs, and the state of your infrastructure, so answers are grounded in your actual environment rather than generic documentation. Whether you're debugging a failed run, exploring a configuration change, or trying to understand what a proposed change will do, the assistant has the context to be useful.

Spacelift Intent

Spacelift Intent is AI-powered infrastructure provisioning through natural language. Rather than asking a question and receiving an answer or getting generated Terraform code, you describe what you want your infrastructure to do. Intent interprets that, translates it into the appropriate configuration, and acts on it. No code involved at all. Intent talks directly to the providers and deploys what you need, when you need it.

• “I need an S3 bucket to store some temporary files. Adhere to any company policies.”

• “Please create and test a policy to not allow public ec2 instances.”

• “Please build 3 ec2 instances in us-west-1 so I can test a prototype.”

You speak it, and Spacelift Intent builds it. No code, no repository, no waiting. You can use Spacelift Intent by chatting with Infra assistant or have your developers connect their agent or LLM of choice to our MCP server.

Getting Started

Spacelift Intelligence is available now. For more information, check out the blog here. Also, sign up for the upcoming webinar where we will discuss Spacelift Intelligence in depth here, and see it in action here!