Self-hosted Google Cloud Run MCP alternatives
The Cloud Run server runs locally, so the process and your Google Cloud credentials stay on your own machine. The agent connects over stdio; the server then calls Google Cloud APIs to deploy and inspect services. Running it yourself controls where the credential lives, not where the deploy actually lands.
Every server below installs and runs locally the same way. They authenticate to their own cloud or database, so what self-hosting keeps on your infrastructure is the process and the keys, while the deploys and queries still reach each provider's API.
The 8 best self-hosted alternatives
The AWS Labs server runs locally and executes any AWS CLI command with validation, read-only mode, and suggestions, keeping your AWS credentials in your own process.
Set up AWS (AWS Labs) →- AzureOfficial
Microsoft's Azure server installs locally to manage 40+ services, storage, Key Vault, Cosmos DB, SQL, Monitor, AKS, with the credential held on your machine rather than a hosted endpoint.
Set up Azure → - DigitalOceanOfficial
Run locally, DigitalOcean's server covers App Platform, Droplets, managed databases, Kubernetes, networking, and storage from a process you control.
Set up DigitalOcean → - HerokuOfficial
Heroku's Platform server runs locally to deploy apps, scale dynos, manage add-ons and pipelines, and operate Postgres, the closest deploy-and-operate match you can self-host.
Set up Heroku → - NetlifyOfficial
Installed locally to create, deploy, and manage projects, env vars, forms, and access controls, the Netlify server keeps the deploy credential on your own infrastructure.
Set up Netlify → Operating a deployment means watching it. Grafana Labs' server runs locally to query dashboards, Prometheus, Loki, incidents, alerts, and OnCall, pairing observability with a self-hosted deploy stack.
Set up Grafana →Reading local state, the DBHub SQLite server runs against a database file with execute_sql and search_objects, a zero-dependency, self-contained option for an app's data.
Set up SQLite (DBHub) →DBHub connects an agent to Postgres via a DSN through execute_sql and search_objects, running locally so database credentials stay in your own process beside the deploy server.
Set up DBHub (Postgres) →
How to choose
All of these run as local commands, keeping the process and credentials on your infrastructure. Heroku is the closest deploy-and-operate match to Cloud Run; AWS, Azure, and DigitalOcean are broader cloud consoles. Grafana adds observability, and the DBHub SQLite and Postgres servers cover an app's data. The deploys and queries still reach each provider's API; only the keys stay home.
FAQ
- Can the Google Cloud Run MCP server be self-hosted?
- Yes. It runs as a local command over stdio, holding your Google Cloud credentials on your own machine. It still calls Google Cloud APIs to deploy and inspect services, so self-hosting protects the credential and process rather than the deploy path.
- Which self-hosted server is closest to Cloud Run for deploys?
- Heroku's Platform server, since it deploys apps and operates services much like the Cloud Run server. For broad cloud management run locally, the AWS, Azure, and DigitalOcean servers cover far more services from your own process.