add-ml-service

What problem does it solve?

It helps teams add a new LiftWing inference service to Wikimedia’s KServe deployment setup without hand-editing fragile helmfile and values files.

Core Features & Use Cases

• Adds to an existing namespace: appends a new inference_services entry to the correct values.yaml and staging override file(s).
• Creates a brand-new namespace scaffold: generates helmfile.yaml, values.yaml, and values-ml-staging-codfw.yaml (and optional eqiad-only env support) from the repo’s templates.
• Configures operational essentials: collects and wires CPU/memory/GPU requests/limits, replica ranges, optional request batching parameters, model image naming, environment variables, and network egress endpoints for new namespaces.

Quick Start

Tell the skill the new inference service name (kebab-case) when prompted so it can either append the service to an existing namespace or scaffold a new one end-to-end.