Zilliz Cloud MCP server

What the model can do

Zilliz Cloud managed Milvus vector database: projects, clusters with plan, size, status, region and endpoint, a cluster's metrics over time (compute and storage use, request counts, latency, failure rates, entities), backups, alert rules, control-plane jobs, and a cluster's collections with their schema, row count and load state; when you ask, can suspend, resume, resize or drop a cluster, create, delete or restore backups, create or delete alert rules, and load, release or drop a collection.

Reads

• zilliz_list_projectsZilliz Cloud projects the key can see: id, name, plan, regions and how many clusters.
• zilliz_list_clustersZilliz Cloud clusters: id, name, status (RUNNING, SUSPENDED…), plan (Free, Serverless, Standard, Enterprise…), compute units, region…
• zilliz_get_clusterOne Zilliz Cloud cluster in detail: status, plan, compute units and type, replicas, storage size, snapshots, autoscaling…
• zilliz_cluster_metricsA Zilliz Cloud cluster's metrics over a period: compute (CU_COMPUTATION) and capacity use, storage, read and write vCUs, request counts…
• zilliz_list_backupsZilliz Cloud backups (dedicated clusters only) that are available or being created: id, name, cluster, CLUSTER or COLLECTION, manual or automatic…
• zilliz_list_alert_rulesA Zilliz Cloud project's alert rules: name, enabled, level, metric, comparison and threshold, window, which clusters, and how it notifies…
• zilliz_get_jobA Zilliz Cloud control-plane job (a backup, restore or migration started by an action): type, status, progress, start, end and failure reason.
• zilliz_collectionsA Zilliz Cloud cluster's collections, read from the cluster's own endpoint.

Acts, when you ask

These change something in Zilliz Cloud. The model is told to use them only when you ask for exactly that, and each one is granted by name: per team on a shared connection, by the owner on a personal one.

• zilliz_suspend_clusterSuspends a dedicated Zilliz Cloud cluster.
• zilliz_resume_clusterResumes a suspended dedicated Zilliz Cloud cluster; compute billing restarts and it takes several minutes to reach RUNNING.
• zilliz_resize_clusterChanges a dedicated Zilliz Cloud cluster's compute units (1 to 1024) and/or replicas (1 to 10).
• zilliz_drop_clusterDrops a Zilliz Cloud cluster and everything in it.
• zilliz_create_backupStarts a backup of a dedicated Zilliz Cloud cluster, whole (CLUSTER) or named collections (COLLECTION); backup storage is billed.
• zilliz_delete_backupPermanently deletes a Zilliz Cloud backup; it can no longer be restored.
• zilliz_restore_backupRestores a cluster backup into a NEW dedicated Zilliz Cloud cluster in the same region and CU type, which Zilliz then bills.
• zilliz_create_alert_ruleCreates an alert rule in a Zilliz Cloud project that emails the named members and/or everyone holding the named organization or project roles when a…
• zilliz_delete_alert_rulePermanently deletes a Zilliz Cloud alert rule; nobody is alerted by it again.
• zilliz_load_collectionLoads a Zilliz Cloud collection into memory so it can be searched and queried (it uses cluster memory).
• zilliz_release_collectionReleases a Zilliz Cloud collection from memory; searches and queries on it fail until it is loaded again.
• zilliz_drop_collectionDrops a Zilliz Cloud collection and permanently erases all of its data; it cannot be undone except from a backup.

Connect it

First connect Zilliz Cloud on Dashboard → Connections, then add Memtro to your client once. The Zilliz Cloud tools appear alongside the tools of everything else you have connected; nothing is configured per connector.

claude plugin marketplace add memtro-org/claude-memtro
claude plugin install memtro@memtro --config api_key=sk-memtro-...

{
  "mcpServers": {
    "memtro": {
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

{
  "mcp": {
    "memtro": {
      "type": "remote",
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

Who may use it

The key identifies the person, and the server offers that person the tools they may use everywhere else in Memtro. A shared Zilliz Cloud connection is used by the teams an administrator granted it to, tool by tool, on the Permissions page; a personal connection is used by its owner alone, and its tools that act stay off until the owner allows them. See Teams and permissions and Connectors.

Ask it like this

• › Which clusters had the highest p99 latency today?
• › Is the products collection loaded, and how many rows does it hold?
• › Which backups expire this week?
• › What alert rules does the search project have?
• › Suspend the staging cluster.

Questions

Is this an official Zilliz Cloud MCP server?

No. It is Memtro's MCP server with Memtro's Zilliz Cloud connector behind it: the connector reads Zilliz Cloud through its API with the credentials you connect on Dashboard → Connections, and Zilliz Cloud publishes nothing here. The same endpoint carries every other system you connect and Memtro's memory tools.

What does it cost?

Nothing on top of Memtro. One price per organisation by its number of users, from $50 a month for up to 10 users after a 7-day free trial, with every connector, the MCP server, jobs and memory included. You bring your own model keys.

Can it change anything in Zilliz Cloud?

Only when you ask, and only what it has been granted: suspend a dedicated cluster (stops compute billing, takes it offline); resume a suspended cluster (compute billing restarts); resize a cluster's compute units or replicas (changes the bill); drop a cluster (recycle bin for 30 days, free clusters excepted); back up a cluster or some of its collections (storage is billed); delete a backup; restore a backup to a new cluster (a new billed cluster); create an alert rule that emails named members or roles; delete an alert rule; load a collection into memory; release a collection from memory; drop a collection and all its data. Every other tool reads, searches or lists. An administrator grants each acting tool by name per team, and a personal connection acts only after its owner allows it.

Which clients work?

Any MCP client that speaks Streamable HTTP: Claude Code, Cursor, Claude Desktop, OpenCode, Codex, Zed, Windsurf, Cline and the Anthropic and OpenAI SDKs. There is no local process to run; the server is remote and authenticated with your key.

Other MCP servers · Databases and warehouses

• Postgres MCP server
• MySQL and MariaDB MCP server
• SQL Server MCP server
• BigQuery MCP server
• Snowflake MCP server
• AWS Athena MCP server