Terraform MCP server: general availability for HCP Terraform and Terraform Enterprise

2026-06-11 · Publisher : HashiCorp (IBM) · Domain : Cloud & DevOps

On 11 June 2026, HashiCorp declares its Terraform MCP server generally available for HCP Terraform and Terraform Enterprise: code generation, plan analysis and workspace queries.

### Background On 11 June 2026, HashiCorp, an IBM company, announces general availability of the Terraform MCP server, for HCP Terraform and Terraform Enterprise. The Model Context Protocol lets AI assistants (GitHub Copilot, IBM Bob, Claude and other compatible tools) interact with Terraform in natural language. ### Key facts - Code generation from modules approved by the organization in its private registries. - Workspace insights, for example "which workspaces haven't been updated in 90 days?". - Plan analysis: natural-language explanation of planned infrastructure changes. - "Security by design": assistants receive only the necessary metadata, credentials stay in the deployment environment. - Controls: CORS policies, rate limiting, OpenTelemetry integration for auditing; local or shared-service deployment, with personal Terraform tokens to keep access individual. ### Why it matters General availability makes MCP a supported channel for driving infrastructure as code with agents, beyond experimentation. For platform teams, the challenge is to control what the assistant sees and can do: the personal token model and audit logs provide the framework. For consulting firms, it is a building block to assess in provisioning chains, with the same review rules as any infrastructure change. As with any agent access, the rights of the Terraform tokens used should be reviewed and audit logs kept. ### Takeaway Infrastructure as code opens to AI assistants through an official channel, with the platform's access controls.

Tags : MCP, Agents, Infrastructure as Code, DevOps, IA