Zilliz MCP Server

List all projects scoped to API Key in Zilliz Cloud. Args: None Returns: JSON string containing the API response with projects data Example: '[{"project_name": "Default Project", "project_id": "proj-f5b02814db7ccfe2d16293", "instance_count": 0, "create_time": "2023-06-14T06:59:07Z"}]'

List all clusters scoped to API Key in Zilliz Cloud. If you want to list all clusters, you can set page_size to 100 and current_page to 1. Args: page_size: The number of records to include in each response (default: 10) current_page: The current page number (default: 1) Returns: List containing cluster data Example: [ { "cluster_id": "inxx-xxxxxxxxxxxxxxx", "cluster_name": "dedicated-3", "description": "", "region_id": "aws-us-west-2", "plan": "Standard", "cu_type": "Performance-optimized", "cu_size": 1, "status": "RUNNING", "connect_address": "https://inxx-xxxxxxxxxxxxxxx.aws-us-west-2.vectordb.zillizcloud.com:19530", "private_link_address": "", "project_id": "proj-xxxxxxxxxxxxxxxxxxxxxx", "create_time": "2024-06-30T16:49:50Z" } ]

Create a free cluster in Zilliz Cloud. Args: cluster_name: Name of the cluster to create project_id: ID of the project to which the cluster belongs Returns: Dict containing cluster creation info Example: { "cluster_id": "inxx-xxxxxxxxxxxxxxx", "username": "db_xxxxxxxx", "prompt": "successfully submitted, cluster is being created..." }

Describe a cluster in detail. Args: cluster_id: ID of the cluster whose details are to return Returns: Dict containing detailed cluster information Example: { "cluster_id": "inxx-xxxxxxxxxxxxxxx", "cluster_name": "Free-01", "project_id": "proj-b44a39b0c51cf21791a841", "description": "", "region_id": "gcp-us-west1", "cu_type": "", "plan": "Free", "status": "RUNNING", "connect_address": "https://inxx-xxxxxxxxxxxxxxx.api.gcp-us-west1.zillizcloud.com", "private_link_address": "", "cu_size": 0, "storage_size": 0, "snapshot_number": 0, "create_progress": 100, "create_time": "2024-06-24T12:35:09Z" }

Suspend a dedicated cluster in Zilliz Cloud. Args: cluster_id: ID of the cluster to suspend Returns: Dict containing cluster suspension info Example: { "cluster_id": "inxx-xxxxxxxxxxxxxxx", "prompt": "Successfully Submitted. The cluster will not incur any computing costs when suspended. You will only be billed for the storage costs during this time." }

Resume a dedicated cluster in Zilliz Cloud. Args: cluster_id: ID of the cluster to resume Returns: Dict containing cluster resumption info Example: { "cluster_id": "inxx-xxxxxxxxxxxxxxx", "prompt": "successfully Submitted. Cluster is being resumed, which is expected to takes several minutes. You can access data about the creation progress and status of your cluster by DescribeCluster API. Once the cluster status is RUNNING, you may access your vector database using the SDK." }

Query the metrics of a specific cluster. Args: cluster_id: ID of the target cluster start: Starting date and time in ISO 8601 timestamp format (optional, use with end) end: Ending date and time in ISO 8601 timestamp format (optional, use with start) period: Duration in ISO 8601 duration format (optional, use when start/end not set) granularity: Time interval for metrics reporting in ISO 8601 duration format (minimum PT30S) metric_queries: List of metric queries, each containing 'metricName' and 'stat' fields - metricName: Name of the metric. Available options: * CU_COMPUTATION - Compute unit computation usage * CU_CAPACITY - Compute unit capacity * STORAGE_USE - Storage usage * REQ_INSERT_COUNT - Insert request count * REQ_BULK_INSERT_COUNT - Bulk insert request count * REQ_UPSERT_COUNT - Upsert request count * REQ_DELETE_COUNT - Delete request count * REQ_SEARCH_COUNT - Search request count * REQ_QUERY_COUNT - Query request count * VECTOR_REQ_INSERT_COUNT - Vector insert request count * VECTOR_REQ_UPSERT_COUNT - Vector upsert request count * VECTOR_REQ_SEARCH_COUNT - Vector search request count * REQ_INSERT_LATENCY_P99 - Insert request latency P99 * REQ_BULK_INSERT_LATENCY_P99 - Bulk insert request latency P99 * REQ_UPSERT_LATENCY_P99 - Upsert request latency P99 * REQ_DELETE_LATENCY_P99 - Delete request latency P99 * REQ_SEARCH_LATENCY_P99 - Search request latency P99 * REQ_QUERY_LATENCY_P99 - Query request latency P99 * REQ_SUCCESS_RATE - Request success rate * REQ_FAIL_RATE - Request failure rate * REQ_FAIL_RATE_INSERT - Insert request failure rate * REQ_FAIL_RATE_BULK_INSERT - Bulk insert request failure rate * REQ_FAIL_RATE_UPSERT - Upsert request failure rate * REQ_FAIL_RATE_DELETE - Delete request failure rate * REQ_FAIL_RATE_SEARCH - Search request failure rate * REQ_FAIL_RATE_QUERY - Query request failure rate * ENTITIES_LOADED - Number of loaded entities * ENTITIES_INSERT_RATE - Entity insert rate * COLLECTIONS_COUNT - Collection count * ENTITIES_COUNT - Total entity count - stat: Statistical method (AVG for average, P99 for 99th percentile - P99 only valid for latency metrics) Returns: Dict containing cluster metrics data Example: { "code": 0, "data": { "results": [ { "name": "CU_COMPUTATION", "stat": "AVG", "unit": "percent", "values": [ { "timestamp": "2024-06-30T16:09:53Z", "value": "1.00" } ] } ] } }

List all databases in the current cluster. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field Returns: List of database names Example: [ "default", "test" ]

List all collection names in the specified database. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field db_name: The name of an existing database. Pass explicit dbName or leave empty when cluster is free or serverless Returns: JSON string containing list of collection names Example: '["quick_setup_new", "customized_setup_1", "customized_setup_2"]' If no collections found, returns: '[]'

Create a collection in a specified cluster using Quick Setup. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field collection_name: The name of the collection to create dimension: The number of dimensions a vector value should have db_name: The name of the database. Pass explicit dbName or leave empty when cluster is free or serverless metric_type: The metric type (default: "COSINE", options: "L2", "IP", "COSINE") Ask the user to select the metric type, if user does not select, use default value "COSINE" id_type: The data type of the primary field (default: "Int64", options: "Int64", "VarChar") auto_id: Whether the primary field automatically increments (default: True) primary_field_name: The name of the primary field (default: "id") vector_field_name: The name of the vector field (default: "vector") Returns: Dict containing the response Example: { "code": 0, "data": {} }

Describe the details of a collection. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field collection_name: The name of the collection to describe db_name: The name of the database. Pass explicit dbName or leave empty when cluster is free or serverless Returns: Dict containing detailed information about the specified collection Example: { "code": 0, "data": { "aliases": [], "autoId": false, "collectionID": 448707763883002000, "collectionName": "test_collection", "consistencyLevel": "Bounded", "description": "", "enableDynamicField": true, "fields": [ { "autoId": false, "description": "", "id": 100, "name": "id", "partitionKey": false, "primaryKey": true, "type": "Int64" }, { "autoId": false, "description": "", "id": 101, "name": "vector", "params": [ { "key": "dim", "value": "5" } ], "partitionKey": false, "primaryKey": false, "type": "FloatVector" } ], "indexes": [ { "fieldName": "vector", "indexName": "vector", "metricType": "COSINE" } ], "load": "LoadStateLoaded", "partitionsNum": 1, "properties": [] } }

Insert data into a specific collection. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field collection_name: The name of an existing collection data: An entity object or an array of entity objects. Note that the keys in an entity object should match the collection schema db_name: The name of the target database. Pass explicit dbName or leave empty when cluster is free or serverless Returns: Dict containing the response with insert count and insert IDs Example: { "code": 0, "data": { "insertCount": 10, "insertIds": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] } }

Delete entities from a collection by filtering conditions or primary keys. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field collection_name: The name of an existing collection filter: A scalar filtering condition to filter matching entities. You can set this parameter to an empty string to skip scalar filtering. To build a scalar filtering condition, refer to Reference on Scalar Filters db_name: The name of the target database. Pass explicit dbName or leave empty when cluster is free or serverless partition_name: The name of a partition in the current collection. If specified, the data is to be deleted from the specified partition Returns: Dict containing the response Example: { "code": 0, "cost": 0, "data": {} }

Conduct a vector similarity search with an optional scalar filtering expression. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field collection_name: The name of the collection to which this operation applies data: A list of vector embeddings. Zilliz Cloud searches for the most similar vector embeddings to the specified ones anns_field: The name of the vector field limit: The total number of entities to return (default: 10). The sum of this value and offset should be less than 100 db_name: The name of the database. Pass explicit dbName or leave empty when cluster is free or serverless filter: The filter used to find matches for the search offset: The number of records to skip in the search result. The sum of this value and limit should be less than 16,384 grouping_field: The name of the field that serves as the aggregation criteria output_fields: An array of fields to return along with the search results metric_type: The name of the metric type that applies to the current search (L2, IP, COSINE) search_params: Extra search parameters including radius and range_filter partition_names: The name of the partitions to which this operation applies consistency_level: The consistency level of the search operation (Strong, Eventually, Bounded) Returns: Dict containing the search results Example: { "code": 0, "data": [ { "color": "orange_6781", "distance": 1, "id": 448300048035776800 }, { "color": "red_4794", "distance": 0.9353201, "id": 448300048035776800 } ] }

Conduct a filtering on the scalar field with a specified boolean expression. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field collection_name: The name of the collection to which this operation applies filter: The filter used to find matches for the search db_name: The name of the database. Pass explicit dbName or leave empty when cluster is free or serverless output_fields: An array of fields to return along with the query results partition_names: The name of the partitions to which this operation applies. If not set, the operation applies to all partitions in the collection limit: The total number of entities to return (default: 100). The sum of this value and offset should be less than 16,384 offset: The number of records to skip in the search result. The sum of this value and limit should be less than 16,384 Returns: Dict containing the query results Example: { "code": 0, "cost": 0, "data": [ { "color": "red_7025", "id": 1 }, { "color": "red_4794", "id": 4 }, { "color": "red_9392", "id": 6 } ] }

Search for entities based on vector similarity and scalar filtering and rerank the results using a specified strategy. Args: cluster_id: ID of the cluster region_id: ID of the cloud region hosting the cluster endpoint: The cluster endpoint URL. Can be obtained by calling describe_cluster and using the connect_address field collection_name: The name of the collection to which this operation applies search_requests: List of search parameters for different vector fields. Each search request should contain: - data: A list of vector embeddings - annsField: The name of the vector field - filter: A boolean expression filter (optional) - groupingField: The name of the field that serve as the aggregation criteria (optional) - metricType: The metric type (L2, IP, COSINE) (optional) - limit: The number of entities to return - offset: The number of entities to skip (optional, default: 0) - ignoreGrowing: Whether to ignore entities in growing segments (optional, default: false) - params: Extra search parameters with radius and range_filter (optional) rerank_strategy: The name of the reranking strategy (rrf, weighted) rerank_params: Parameters related to the specified strategy (e.g., {"k": 10} for rrf) limit: The total number of entities to return. The sum of this value and offset should be less than 16,384 db_name: The name of the database. Pass explicit dbName or leave empty when cluster is free or serverless partition_names: The name of the partitions to which this operation applies output_fields: An array of fields to return along with the search results consistency_level: The consistency level of the search operation (Strong, Eventually, Bounded) Returns: Dict containing the hybrid search results Example: { "code": 0, "cost": 0, "data": [ { "book_describe": "book_105", "distance": 0.09090909, "id": 450519760774180800, "user_id": 5, "word_count": 105 } ] }