Connect the IcePanel MCP server to Unblocked to give AI agents live architecture context

An agent working inside one repository can’t see how that service fits into the wider system — which domain it belongs to, what it talks to, which team owns it, and which architecture decisions constrain it. IcePanel’s MCP server exposes your architecture model and decision records to AI agents, and Unblocked connects to it once so every agent can ask about your architecture without its own IcePanel setup.

This is a third-party MCP server: IcePanel hosts and maintains it, and Unblocked connects to it as a verified MCP server. Nothing is ingested or indexed — Unblocked calls the read-only tools you select at the moment a question needs them, so answers reflect your landscape as it is modelled now.

Wiring IcePanel’s MCP server straight into an agent gives it model objects and connections. Unblocked’s Context Engine reasons across those results and your other sources — following a connection in the model to the service that implements it, the pull request that changed it, and the Slack thread where the decision record was debated — and returns one cited answer, including where the model and the code disagree.

Once connected, any MCP-compatible coding agent — Claude Code, Cursor, Devin, GitHub Copilot, Codex — reaches IcePanel context through a single Unblocked MCP server connection.

What you get

• Your architecture model, live — Agents query model objects, domains, connections, diagrams, and flows as they are modelled in IcePanel right now, not a snapshot exported months ago.
• Decision records in reach — Agents can read your architecture decision records, along with the teams, technology choices, and tags attached to the model, before they propose a change.
• Read-only, even with write access enabled — IcePanel can allow writes to model objects, connections, and decision records. Unblocked offers only its read-only tools, and you choose which of them it may call from Settings → Data Sources.
• Model checked against the code — The Context Engine reconciles IcePanel’s model with your code, pull requests, and conversations, so a connection that no longer exists in code — or a new dependency missing from the model — comes to light with citations.

Prerequisites

• An Unblocked account (free to start).
• An IcePanel organization on a paid plan, and an IcePanel editor or admin role to authorize OAuth access to the landscape you want to query.
• An MCP-compatible coding agent (Claude Code, Cursor, Devin, etc.) to use the context in your editor.

Step 1: Connect IcePanel to Unblocked

1. In Unblocked, go to Settings → Data Sources → Connect another data source, and under Documentation select IcePanel.
2. The IcePanel MCP server URL is filled in for you. Choose OAuth, click Connect, and sign in to IcePanel to authorize Unblocked.
3. Once Unblocked discovers the server’s tools, select the read-only tools it may use and click Save Settings.

Step 2: Install the Unblocked MCP server

On macOS or Linux, one command auto-detects and configures the supported agents on your machine:

On Windows, or when you prefer remote MCP, add the server manually using the configuration below, or follow the manual install guide .

claude mcp add -s user --transport http unblocked https://getunblocked.com/api/mcpsse

{
  "mcpServers": {
    "unblocked": {
      "url": "https://getunblocked.com/api/mcpsse"
    }
  }
}

https://getunblocked.com/api/mcpsse

To give Claude Code, Cursor, GitHub Copilot, Codex, or another MCP-compatible AI coding agent access to IcePanel context, connect IcePanel to Unblocked and add the single Unblocked MCP server to your agent.

Unblocked acts as the MCP context layer for your IcePanel data, searching architecture diagrams and models alongside your other connected sources and returning cited results through these tools:

• context_researchResearches a question across your connected sources and calls the IcePanel tools you selected when your architecture model, diagrams, or decision records are relevant, returning a synthesized answer with citations.
• context_get_urlsFetches the full content behind URLs your agent encounters across connected sources — Jira, Linear, Notion, Confluence, Slack, GitHub, and more.

Questions you (or your agent) can ask Unblocked about your IcePanel data:

• Which services does the payments domain depend on in IcePanel, and which of those connections changed in recent pull requests?
• What does team Platform own according to our architecture model, and who has been changing those services lately?
• Is there a decision record about moving off the shared database, and what did the Slack discussion around it conclude?
• Does the checkout flow modelled in IcePanel still match the calls the code makes today?