Skip to main content
The StreamNative MCP Server organizes tools into tool groups. Each feature ID enables a group of related tools, not necessarily one MCP tool. Pulsar, Kafka, and dynamic feature IDs apply to both the Remote MCP Server and the local MCP Server, unless a note states otherwise. StreamNative Cloud organization feature IDs apply to Remote MCP organization-level sessions. Mixed administration groups use separate read and write tools. For example, kafka-admin can expose kafka_admin_topics_read for inspection and kafka_admin_topics_write for mutations. Read tools are marked as read-only and support safe inspection operations. Write tools are marked as mutating or destructive and can create, update, delete, apply, trigger, produce, or otherwise change resources. In read-only mode, the server omits write or destructive tools where possible and also blocks mutation calls server-side.

​
StreamNative Cloud tool groups

Organization-level sessions expose StreamNative Cloud tools for cluster discovery and control-plane resources. For workflow details, see Organization-Level Tools. Read tools support list and get. Write tools support apply and delete. Write tools accept manifests as JSON strings and support dry_run=true for validation. Use sncloud_resource_catalog to discover supported kinds, then use sncloud_resource_schema to get summary, example, runtime JSON schema, or selected schema paths before writing manifests. After sncloud_context_use_cluster selects a cluster in a root or organization-level session, sncloud_logs is also available for logs from Functions, Sources, Sinks, and Kafka Connect connectors in the selected cluster. sncloud_logs is cluster-scoped and appears only after cluster selection.

​
Pulsar tool groups

Pulsar clusters support the following tool groups.

​
Kafka tool groups

Kafka clusters support the following tool groups.
Kafka topic, partition, and consumer-group tools are enabled by kafka-admin. There are no separate Remote MCP feature IDs named kafka-admin-topics, kafka-admin-partitions, or kafka-admin-groups.

​
Combination shortcuts

Use these shortcut IDs to enable multiple tool groups at once.

​
Dynamic tools

In addition to static tool groups, the Remote MCP Server supports dynamic tools that expose your own workloads as MCP tools.
Dynamic tools are available on fixed cluster endpoints. They are not added after selecting a cluster inside an organization-level session with sncloud_context_use_cluster.

​
Client tool parameter notes

  • Kafka consume tools use group for the Kafka consumer group ID. When group is omitted, the server uses an ephemeral group and cannot read committed offsets.
  • Pulsar consume tools use subscription-name for the subscription. This is the Pulsar equivalent of a durable cursor; do not pass a Kafka group parameter to Pulsar tools.
  • StreamNative Cloud domain read tools use operation=list|get, resource, and optional name. name is required for get.
  • StreamNative Cloud domain write tools use operation=apply|delete, resource, and either manifest for apply or name for delete.
  • StreamNative Cloud write tools use manifest as a JSON string. Do not pass YAML or an object. Use dry_run=true for apply before applying changes.
  • sncloud_logs requires a selected cluster in a root or organization-level session. Use it for logs from Functions, Sources, Sinks, and Kafka Connect connectors.
  • StreamNative Cloud read tools omit status and metadata.managedFields unless you set includeStatus=true or includeManagedFields=true.
  • sncloud_resource_schema defaults to format=summary. Use format=example for an example manifest, format=jsonschema for runtime Kubernetes schema when available, and paths such as spec.broker or spec.clusterRefs to reduce schema size.

​
Usage tips

  • Start minimal. Enable only the tool groups you need. Fewer tools reduce token usage and help AI agents select the right tool more reliably.
  • Use shortcuts for broad access. If you need all Pulsar tools, use all-pulsar instead of listing each group individually.
  • Match groups to your cluster type. Pulsar tool groups only work on Pulsar clusters, and Kafka tool groups only work on Kafka clusters. Enabling mismatched groups has no effect.
  • Combine groups freely. You can mix StreamNative Cloud, Pulsar, Kafka, and dynamic tool groups. For example, sncloud_context,pulsar-admin-topics,pulsar-client enables cluster discovery, Pulsar topic administration, and Pulsar message operations.
  • Use StreamNative Cloud schema helpers before writes. Call sncloud_resource_catalog, then sncloud_resource_schema, then a read tool, then a write tool with dry_run=true before applying changes.