Article · part of a guide
Which Terraform Platform Has the Best MCP Server?
HashiCorp, Scalr, Spacelift, and env0 all ship MCP servers now, and they differ most on one axis: how much the agent is allowed to change. A comparison for teams deciding which platform fits agent-driven workflows.
Key takeaways
- As of November 2026 every major Terraform platform except Atlantis ships an MCP server. The differentiating question isn't whether one exists but what it can read, what it can write, and how it authenticates.
- Spacelift's MCP wraps its GraphQL API including mutations and Intent-managed resource provisioning; env0's can deploy and approve environments. Both suit teams that want agents executing changes end to end.
- As of November 2026, Scalr's MCP server is read-mostly: its only write operations are creating workspaces and variables, and it can't trigger runs or approve applies. Scalr admins choose which tools the hosted server offers, one tool at a time.
- Scalr and Spacelift offer hosted remote MCP endpoints with per-user OAuth that inherits platform RBAC. HashiCorp's hosted endpoint is in beta, takes an HCP Terraform API token, and serves US HCP Terraform organizations only; env0's server runs locally with an API key pair.
- Vendor AI assistants are gated differently as of November 2026: Spacelift's run summaries come through Infra Assistant on all plans, HashiCorp's Infragraph is in limited availability for Standard and Premium US customers, and env0 listed Cloud Analyst with its top Cloud Pilot tier in July 2026. Scalr AI, which explains errored runs, is included in paid Scalr plans.
HashiCorp, Scalr, Spacelift, and env0 all ship an MCP server as of November 2026, so the question is what the agent can change through it, and whose identity it uses when it does. If you want agents executing changes end to end, Spacelift and env0 expose the most authority (runs, approvals, and in Spacelift's case resource provisioning through Intent). If you want agents investigating while humans hold the apply, Scalr's read-mostly server with per-user OAuth fits that today, since it can't trigger runs or approve applies. HashiCorp's is the best registry-docs source if you're staying on HCP Terraform.
Scalr is our product, and this comparison reflects a view we hold about agent access. We've tried to keep every factual claim dated and checkable, and where a competitor's design is the better fit for a use case, we say so.
What is a Terraform MCP server?
A Terraform MCP server is a Model Context Protocol endpoint that exposes a Terraform platform's data and operations as tools an AI coding agent can call directly, instead of the agent shelling out to the CLI or scraping a web UI. In practice that means an assistant can ask for a workspace, read a plan, look up provider documentation, or open a run, using the platform's own authentication and permissions rather than a copied credential.
What Should You Compare MCP Servers On?
Tool counts matter less than four properties that determine what a server means for your risk posture, drawn from the access-class taxonomy in our primer on IaC with AI coding agents:
- Read scope. Registry documentation only, or control-plane data: workspaces, runs, logs, variables, drift, state-derived inventory.
- Write scope. Can it mutate configuration objects? Trigger runs? Approve applies? Provision resources directly? Each step widens what a compromised or confused session can do.
- Identity model. A hosted endpoint with per-user OAuth inherits your existing RBAC per person. A local server with a static token is exactly as safe as that token's scope and storage.
- Blast-radius controls. Toolset flags, read/write scope separation, and whether destructive operations need explicit enablement.
How Do the Platforms Compare?
The snapshot below was verified against vendor documentation and repositories on November 13, 2026. These servers change monthly; treat it as a starting point, not a purchase decision.
| Platform | Read scope | Write scope | Hosted remote + OAuth |
|---|---|---|---|
| HashiCorp terraform-mcp-server (v1.3.0) | Registry docs plus workspaces, runs, plan JSON, logs, state versions | Workspaces, variables, projects, teams; runs; applies (self-run only, behind a flag) | Hosted beta with an API token, US HCP Terraform only; or self-run |
| Scalr | Whole control plane incl. IAM, billing, usage (52 tools) | Workspaces and variables only; no runs, no approvals | Yes, per-user OAuth; admins allow or deny each tool |
| Spacelift | GraphQL read, provider schemas, policy simulation | GraphQL mutations, plus Intent-managed resource provisioning | Yes, per-user OAuth |
| env0 | Projects, environments, plan logs, error analysis, deployment history, cloud inventory | Deploy, approve, cancel, abort; IaC generation | No, local with API keys |
The detail behind the cells: on a self-run HashiCorp server, applies, discards, cancels, deletes, and force-unlocks are gated behind ENABLE_TF_OPERATIONS=true. HashiCorp's hosted server at mcp.terraform.io leaves those operations out entirely, doesn't support Terraform Enterprise or HCP Europe, and has no organization setting to turn it or its toolsets off. Scalr's read scope covers environments, workspaces, runs, logs, policy results, drift, IAM, and billing and usage reports, and its OAuth inherits Scalr RBAC. As of November 2026, Scalr's hosted MCP server lets account admins allow or deny each tool, starts a new integration with every tool denied, and records each change in the audit log. Spacelift's mutations sit under mcp:read and mcp:write scopes. env0 authenticates with an API key pair.
Two entries deserve their own line. Terrakube has an official MCP server with organization, workspace, variable, and module CRUD, but no run execution and no published releases yet, so it's early-stage rather than production tooling. Atlantis has no MCP or AI capability at all; as of November 2026 an MCP server exists only as an unassigned feature request and an unmerged design proposal, which is consistent with Atlantis's position as a lean pull-request automation tool.
Which Design Philosophy Fits Agent Work?
The table shows two philosophies, and naming them matters more than ranking them.
Maximum agent authority. Spacelift and env0 built their servers so an agent can complete a change: trigger the run, approve the deployment, in Spacelift's case create resources through Intent. If your goal is closed-loop automation where agents execute routine changes with humans supervising by exception, that's a real capability, and those platforms deliver more of it. HashiCorp's self-run server lands nearby once operations are enabled, with run creation and apply actions available against HCP Terraform and TFE. Its hosted beta sits in between: an agent can queue a run there but can't apply, discard, or cancel one.
Agent as investigator, human as executor. Scalr's server exposes broad read access, including areas the others mostly don't cover, like billing, IAM, and fleet-wide usage reports, but its write surface stops at creating workspaces and variables. An agent connected to Scalr's MCP can triage a failed run, classify drift, audit policy coverage, or answer "which workspaces still run Terraform 1.5" across the fleet. It can't start a run or approve an apply, because the server doesn't offer those tools today, regardless of what the authenticated user could do in the UI.
We hold the second view. The reviewed plan is the safety boundary. Scalr's docs call run triggering and apply approval "not yet supported," so read today's limit as where the product stands, not a promise. The control that lasts is the admin's: a write tool stays off the hosted server until someone allows it. Prompt injection through tool results is still an unsolved risk; HashiCorp's own MCP security guidance says as much. A server whose worst case is a stray workspace beats one whose worst case is an approved apply.
You might not want that limit, though. If you trust your guardrails, sandboxing, and review pipeline, env0's deploy actions are a feature and human-by-default is friction. That tradeoff decides it, more than any tool count.
What About the Vendors' Own AI Features?
The MCP server is what your agents use; most vendors also sell AI of their own, and the gating differs as of November 2026.
| Vendor | Built-in AI feature | Gating |
|---|---|---|
| Spacelift | Run summaries through Infra Assistant: explains what happened and why a run failed | All plans, token allowance on Free; off by default, admin opt-in |
| env0 | Cloud Analyst: natural-language analytics over deployment data | Listed with the top Cloud Pilot tier as of July 2026; we couldn't re-check it in November |
| HashiCorp | HCP Terraform powered by Infragraph: a unified infrastructure graph meant as the substrate for future agent features | Limited availability since June 2026; Standard plan or above, US only |
| Scalr | Scalr AI (Claude): explains errored runs, summarizes dry runs, and summarizes plans for approvers | No extra charge on paid plans; off by default, account-level opt-in; AI audit log is Enterprise-only |
As of November 2026, Scalr AI uses Claude to explain errored runs and summarize plans for approvers, at no extra charge on paid Scalr plans.
Our post on the Scalr MCP server walks through what that looks like in practice.
If AI-assisted run debugging matters to you, check the gating column as carefully as the feature list.
None of the four vendors documents a plan-tier gate on the MCP server itself, so cost enters through the meter your agent traffic runs against rather than through an MCP line item. Scalr's per-run pricing is the relevant number on our side.
So Which One Is Best?
It depends on what the agent is allowed to do. If you want agents executing changes end to end, env0 and Spacelift expose the most execution authority through MCP, and they've built real controls around it. If you want agents investigating broadly and humans holding the apply, Scalr's read-mostly server with per-user OAuth holds that posture today without extra configuration. HashiCorp's server is the strongest pure registry-documentation source and the natural choice if you're staying on HCP Terraform.
Whichever you pick, connect it read-only first, watch what your agents actually use for a month, and widen scope from evidence rather than from the feature list. Our least-privilege Claude Code setup shows what that looks like on the client side.
Frequently asked questions
What is a Terraform MCP server?
A Terraform MCP server is a Model Context Protocol endpoint that exposes a Terraform platform's data and operations as tools an AI coding agent can call directly, rather than the agent shelling out to the CLI or scraping a web UI. It lets an assistant read workspaces, plans, provider documentation, and run state, and in some implementations open or approve runs, using the platform's own authentication and permissions. As of November 2026 HashiCorp, Scalr, Spacelift, and env0 each publish one; Atlantis does not.
Which Terraform automation platforms have an MCP server as of November 2026?
HashiCorp (terraform-mcp-server v1.3.0, self-run or as a hosted beta, covering registry docs plus HCP Terraform and Terraform Enterprise operations), Scalr (hosted and self-run, 52 tools, read-mostly), Spacelift (hosted, wrapping its GraphQL API with read and write scopes), and env0 (open-source server with deployment actions). Terrakube has an early-stage server without run execution. Atlantis has none; an MCP server exists only as an open feature request and an unmerged design proposal.
Can MCP servers from Terraform platforms apply infrastructure changes?
Some can. HashiCorp's self-run server can create runs and apply, discard, or cancel them when operations are explicitly enabled; its hosted beta can queue runs but leaves out apply, discard, and cancel. Spacelift's mutate tool can change runs, and its Intent tools can create and delete Intent-managed cloud resources. env0's server can deploy, approve, and cancel environments. Scalr's cannot trigger runs or approve applies as of November 2026; its writes are limited to creating workspaces and variables.
Which platform is best for teams adopting AI coding agents?
It depends on how much execution authority you want agents to have. Teams that want agents driving deployments end to end get the most from env0 or Spacelift, whose servers expose deployment and mutation actions. Teams that want agents to investigate, plan, and hand off to a human approver are better matched by Scalr, whose MCP server currently can't trigger runs or approve applies and whose per-user OAuth inherits existing RBAC.
Do these MCP servers cost extra?
None of the four vendors documented a plan-tier restriction on the MCP server itself as of November 2026; HashiCorp's hosted beta is open to every HCP Terraform plan, including Free. The AI assistant features are gated differently: Spacelift's run summaries come through Infra Assistant, available on all plans with a token allowance on Free; HashiCorp's Infragraph requires a Standard plan or above in the US; and env0 listed Cloud Analyst with its top Cloud Pilot tier as of July 2026. Scalr AI is included at no extra charge in paid Scalr plans. Verify against each vendor's current pricing page, since this changes quickly.
About the author

CEO at Scalr
Sebastian Stadil is the CEO at Scalr. He has over 15 years of devops experience, and started his career with AWS in 2004.