InfrOS has introduced its Model Context Protocol (MCP) Server, making its cloud infrastructure design, modernization and validated infrastructure-as-code workflows accessible through compatible AI assistants and development environments. The new offering enables engineering teams to use AI-powered tools to design, manage and evolve cloud infrastructure while streamlining infrastructure-as-code development and validation.
“Infrastructure decisions require more than code that looks right. Teams need designs grounded in their requirements, clear cost-performance trade-offs, and evidence that the infrastructure will behave as intended,” said Naor Porat, co-founder and CEO of InfrOS. “Our MCP Server brings that workflow into their AI assistant from designing new infrastructure to improving what they already run. With InfrOS Evolving Architecture, teams can evaluate changes that boost resilience, reduce costs, improve performance, and adapt their architecture as requirements change in the agentic world. This is a necessity.”
The InfrOS MCP Server allows engineering teams to:
- Develop a design specification. Create projects and versions, update requirements, and identify information still needed before a design run. The tools expose missing requirements and direct the assistant to ask the user for decisions rather than guess.
- Design around the intended environment. Select an existing organizational landing zone or design from scratch, with the option to compare a landing-zone-based design against a greenfield alternative.
- Compare architecture options. Retrieve InfrOS proposals with projected monthly costs in USD, quality ratings, advantages, disadvantages, and the platform’s recommendation. Inspect individual components, including their services, configurations, regions, and costs.
- Generate and retrieve Terraform. Request Infrastructure as Code for your selected architecture for supported proposals, follow its progress, and retrieve the resulting project files once released. The assistant can then place those files in the user’s workspace using its own authorized file-access capabilities.
Authorized organization administrators can also manage landing zones and standing constraints. Where GitHub integration is enabled, they can initiate read-only repository analyses and inspect the infrastructure conventions InfrOS has learned, including Terraform versions, providers, environment structure, and secrets-handling patterns.
Prove Before Deploy: Evidence beyond generated code
For supported architectures and test paths, InfrOS’s Prove Before Deploy capability deploys generated Terraform into its own validation sandbox, not the customer’s production environment and provides evidence across three areas:
- Deployment validation: Check whether the generated infrastructure can be provisioned.
- Connectivity validation: Test supported connections between deployed components.
- Performance benchmarking: Measure supported workload flows, including latency and error rates, against applicable thresholds.
The MCP Server acts on behalf of the signed-in InfrOS user. OAuth authorization separates access to reading, editing, credit-spending operations, and organization administration. Actions remain subject to the user’s permissions and the organization’s enabled features. Users can review connected assistants and revoke their access from their InfrOS profile. Tool-call records capture who acted, which client and tool were used, and the outcome, without retaining the full request and response contents in those records.
Infrastructure delivery remains under the team’s control
The MCP Server makes Terraform files available to the assistant but does not write to customer repositories or open pull requests. Repository changes, code review, and production deployment remain under the team’s control. Some setup and confirmation steps require the InfrOS web application.
Availability
The InfrOS MCP Server is launching in preview for eligible customers. Setup instructions, supported AI clients, and access requirements are available here. Design runs and Terraform generation consume organization credits.
Related News:
InfrOS Evolving Architecture Adapts Cloud Infrastructure for AI Workloads
Black Duck Signal MCP Server Brings Vulnerability Detection to Claude
