Connect Google Cloud with Unblocked to give AI agents live infrastructure context

Answering a question about a running system usually means leaving the editor for the Google Cloud Console. Unblocked connects to Google’s native MCP services so an agent can ask your project directly — what instances exist, what the logs say, what the metrics show — instead of reasoning from a stale screenshot pasted into the conversation.

This is live access, not an index: every request is made against your project at the moment the question is asked, through read-only tools only. Unblocked reconciles what comes back with your code, pull requests, tickets, and incidents, so an agent can move from a log line to the change that produced it.

Access is granted through Workload Identity Federation. Unblocked never holds a service-account key or any other long-lived Google credential, and you choose which of the Compute, Logging, and Monitoring services it may query.

Once connected, any MCP-compatible coding agent — Claude Code, Cursor, Windsurf, GitHub Copilot, Codex — can reach this context through a single Unblocked MCP server connection.

What you get

• Live Compute, Logging, and Monitoring — Agents query your Google Cloud project as it is right now through Google’s native MCP services, rather than an index that was built hours ago.
• Keyless access — Workload Identity Federation trusts one Unblocked AWS role to impersonate a service account you create, so no Google service-account key or other long-lived credential is ever shared.
• Read-only, service-scoped — Only read-only tools are exposed, and only from the services you select — enable Compute, Logging, and Monitoring, or just the one you need.
• Cross-source synthesis — Live cloud answers are combined with your code, pull requests, incidents, and Slack threads, so a symptom in production leads to the change behind it.

Need screenshots or troubleshooting?This is a concise connection summary. Follow the complete setup guide ↗

Prerequisites

• An Unblocked account (free to start).
• A Google Cloud project, and an administrator who can enable APIs, create a service account and Workload Identity Federation resources, and manage project IAM.
• Terraform, the gcloud CLI, or access to the Google Cloud Console to complete the federation setup — all three paths are documented.
• An MCP-compatible coding agent (Claude Code, Cursor, Windsurf, etc.) to use the context in your editor.

Step 1: Connect Google Cloud MCP to Unblocked

1. In Google Cloud, create the service account and Workload Identity Federation resources Unblocked will use, with Terraform, the gcloud CLI, or the Google Cloud Console.
2. Note the two values the setup produces: the service account email and the Workload Identity Provider resource name.
3. In Unblocked, go to Settings → Data Sources → Connect another data source, and select Google Cloud MCP.
4. Paste both values into the connection form and click Connect.
5. Select the Google Cloud MCP services Unblocked can query — Compute, Logging, Monitoring — and click Save Settings.

The Unblocked AWS account and broker role in the setup are fixed values supplied by Unblocked; the Google-side service account, pool, and provider can be renamed to match your conventions.

Step 2: Install the Unblocked MCP server

On macOS or Linux, one command auto-detects and configures your supported agents (Claude Code, Cursor, Windsurf, GitHub Copilot, Codex, OpenCode, VS Code):

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 Google Cloud MCP context, connect Google Cloud MCP to Unblocked and add the single Unblocked MCP server to your agent.

Unblocked acts as the MCP context layer for your Google Cloud MCP data, searching Compute, Cloud Logging, Cloud Monitoring alongside your other connected sources and returning cited results through these tools:

• gcp_mcpLists the read-only tools available from the Google Cloud MCP services you selected, then runs one against your project on request — returning live Compute, Logging, or Monitoring results to the agent.

Questions you (or your agent) can ask Unblocked about your Google Cloud MCP data:

• Which Compute instances are running in this project, and how are they sized?
• What errors did Cloud Logging record for the ingest service in the last hour?
• Is CPU on the batch workers above its usual range this week?
• Which recent pull request could explain the log errors on this service?