tare mcp install

Deploy MCP server pods and ClusterIP services into the cluster
via the embedded mcp-servers Helm chart.

Server selection is an explicit operator choice — pass --servers=<list>
or --servers=all. The CLI uses the embedded catalog seed to derive the set.

After install, paste the printed in-cluster service URLs into the MCP
backend configuration on your TARE dashboard so EAIGW knows how to
route MCP traffic to the new pods.

Examples:
  # Deploy three specific MCP servers
  tare mcp install --servers=jira,slack,confluence

  # Deploy all 15 MCP servers
  tare mcp install --servers=all

  # Override registry (default is registry.tetrate.ai)
  tare mcp install --servers=jira --image-registry=gcr.io/my-mirror

  # Render manifests for GitOps (no cluster contact)
  tare mcp install --servers=jira,slack --print-resources > tars-mcp.yaml

Usage:
  tare mcp install [flags]

Flags:
      --image-registry string   Image registry prefix (set this to your tare mcp images-sync --to destination if you mirrored MCP images). (default "registry.tetrate.ai")
      --namespace string        Namespace to install into. Created if missing. (default "tars-mcp")
      --no-wait                 Don't wait for pods to become ready before returning.
      --print-resources         Render the chart via 'helm template' to stdout and exit. No cluster contact, no confirmation prompt. Pipe to kubectl apply -f - for GitOps workflows.
      --release-name string     Helm release name. (default "tars-mcp")
      --servers string          Comma-separated MCP server aliases to deploy (e.g. jira,slack) or 'all'. Required.
      --yes                     Skip the pre-install confirmation prompt; required on non-TTY runs.

Global Flags:
  -q, --quiet     Suppress progress output; only errors and the final summary go to stderr
  -v, --verbose   Verbose output: plain-mode progress lines plus full per-blob chunk logs to stderr