IceShore

IceShore is a configuration-management platform for teams whose systems have multiplied as AI sped up development. Alongside each system's architecture, it prompts your AI to organize what the system does for the business (its use cases) and serves the result back to it over MCP. Your AI can then explain a change's impact on business operations, on processing, and on external systems itself, instead of guessing from the code alone. The auto-generated architecture diagrams also let you check how well the AI understands each system.

For example, ask your AI "If we stopped the medical-records database, who would be affected?" It reads the use cases behind the architecture and answers with the people and business processes that depend on that database.

Your AI drafts the use cases from your design documents and from infrastructure code such as CloudFormation or Terraform.

No account needed in Claude Code: list_samples, get_content and get_guide work anonymously on the public sample workspaces. In claude.ai and Claude Desktop, adding the connector asks you to sign in (free for a single person). Other tools return 401 with WWW-Authenticate, so your MCP client starts the OAuth sign-in. Once signed in, you can work with your own systems: create and update architecture descriptions and their use cases, upload IaC and design documents, and keep notes.

Before storing code your AI sends, IceShore masks values that look like access keys, passwords, or tokens. How your data is stored and encrypted, and who can access it: https://help.iceshore.ai/en/data-and-trust.html

Endpoint: https://mcp.iceshore.ai/mcp (Streamable HTTP). Auth: OAuth 2.1, authorization server https://api.iceshore.ai, scope adl:rw.