Spinnaker MCP

What GeiserX/spinnaker-mcp does

GeiserX/spinnaker-mcp connects an MCP-compatible AI client to a Spinnaker installation through the Gate API. It exposes tools for discovering and inspecting applications, pipelines, executions, accounts, clusters, server groups, load balancers, firewalls, networks, subnets, images, and instances.

The toolset also covers operational changes. Agents can cancel or pause pipeline executions, delete pipeline and deployment-strategy configurations, inspect orchestration tasks, and evaluate Spring Expression Language expressions against execution context. Pipeline workflows include reading configurations, listing executions, reviewing history, and the broader feature set includes triggering, saving, updating, resuming, and restarting pipeline-related operations.

How it works

The server acts as an MCP-to-Gate bridge. An MCP client sends tool calls, and the server makes corresponding requests to the configured Spinnaker Gate endpoint. Results are returned as JSON for structured Spinnaker data, with raw text used for instance console output.

It can communicate with clients through stdio or HTTP. HTTP listens on 127.0.0.1 by default. The npm package launches the Go binary over stdio, and Docker images and multi-architecture Go binaries are also available.

Setup and configuration

The npm quick-start requires Node.js 18 or newer. Configure an MCP client with the spinnaker-mcp package and provide the Gate URL and token as environment variables:

{
  "mcpServers": {
    "spinnaker": {
      "command": "npx",
      "args": ["-y", "spinnaker-mcp"],
      "env": {
        "GATE_URL": "https://spin-gate.example.com",
        "GATE_TOKEN": "your-token-here"
      }
    }
  }
}

Authentication can use a bearer token, basic authentication, an x509 client certificate, or a Gate session cookie obtained with spinnaker-mcp login. Placeholder-only values such as ${GATE_TOKEN} are treated as unset. Docker, Helm, local builds, HTTP endpoints, and client-specific configuration are also supported by the project.

Tools and capabilities

• Discover applications, accounts, pipelines, clusters, networks, subnets, firewalls, and load balancers.
• Inspect pipeline definitions, revisions, executions, orchestration tasks, server groups, instances, images, and console output.
• Monitor executions and scaling activity, evaluate SpEL, and select a target server group by strategy.
• Pause or cancel executions and remove pipeline or deployment-strategy configurations.
• Work with infrastructure details across configured Spinnaker accounts and regions.
• Use a read-only toolset when only inspection and search operations should be exposed.

Limitations and notes

Deletion of pipelines and deployment strategies is permanent and does not remove past executions. Cancelling an execution stops a running pipeline but does not roll back stages that have already completed. Pausing occurs at the current stage boundary and can be resumed later. SpEL evaluation blocks reflection and class-instantiation patterns. Operations that query infrastructure require the relevant Spinnaker account, region, application, cluster, or resource identifiers.