Qdrant

Get cluster status and information. Returns cluster status, peer information, and raft state for distributed deployments.

Trigger cluster recovery. Attempts to recover the cluster from inconsistent state.

Remove a peer from the cluster. Args: - peerId: ID of the peer to remove - force: Force removal even if unhealthy

Get cluster information for a specific collection. Returns shard distribution and replication status.

Abort an ongoing shard transfer. Args: - collectionName: Name of the collection - shardId: ID of the shard - fromPeerId: Source peer ID - toPeerId: Destination peer ID

Execute multiple point operations in a single request. Supports: upsert, delete, set_payload, overwrite_payload, delete_payload, clear_payload, update_vectors, delete_vectors. More efficient than multiple individual operations. Args: - collectionName: Name of the collection - operations: Array of operations to execute - wait: Wait for all operations to complete

Move a shard from one peer to another. Args: - collectionName: Name of the collection - shardId: ID of the shard to move - fromPeerId: Source peer ID - toPeerId: Destination peer ID - method: Transfer method

Create a replica of a shard on another peer. Args: - collectionName: Name of the collection - shardId: ID of the shard to replicate - fromPeerId: Source peer ID - toPeerId: Destination peer ID for replica

Drop a shard replica from a peer. Args: - collectionName: Name of the collection - shardId: ID of the shard - peerId: Peer ID to drop replica from

Create a custom shard key for a collection. Enables custom sharding for multi-tenant scenarios. Args: - collectionName: Name of the collection - shardKey: Shard key value - shardsNumber: Number of shards for this key - replicationFactor: Replication factor

Delete a custom shard key from a collection. WARNING: This deletes all data in the shard! Args: - collectionName: Name of the collection - shardKey: Shard key to delete

List all collections in the Qdrant database. Returns the names of all available collections.

Get detailed information about a specific collection. Returns collection configuration, status, vector count, and schema information.

Check if a collection exists. Returns true if the collection exists, false otherwise.

Create a new collection in Qdrant. Args: - collectionName: Name for the new collection - vectorSize: Dimension of vectors to store - distance: Distance metric (Cosine, Euclid, Dot, Manhattan) - onDiskPayload: Store payload on disk instead of memory - shardNumber: Number of shards for distributed storage - replicationFactor: Number of replicas for each shard

Create a collection with multiple named vectors. Allows storing multiple vectors per point with different dimensions and metrics. Args: - collectionName: Name for the new collection - vectors: Array of vector configurations, each with name, size, and distance - onDiskPayload: Store payload on disk instead of memory

Update collection parameters. Can modify optimizer settings, HNSW configuration, and other parameters.

Delete a collection and all its data. WARNING: This operation is irreversible!

List all collection aliases. Returns all alias mappings in the database.

Get aliases for a specific collection.

Create an alias for a collection. Aliases provide alternative names to reference collections.

Delete a collection alias.

Rename an existing alias.

Atomically switch an alias from one collection to another. Useful for zero-downtime collection updates.

Create a payload field index for faster filtering. Indexes improve filter performance on specific payload fields. Args: - collectionName: Name of the collection - fieldName: Payload field to index - fieldSchema: Index type (keyword, integer, float, bool, geo, datetime, text, uuid) - wait: Wait for indexing to complete

Create a full-text search index on a payload field. Enables text search with tokenization. Args: - collectionName: Name of the collection - fieldName: Payload field to index - tokenizer: Tokenization strategy - minTokenLen: Minimum token length - maxTokenLen: Maximum token length - lowercase: Convert to lowercase

Create an integer index with lookup and/or range support. Args: - collectionName: Name of the collection - fieldName: Payload field to index - lookup: Enable direct lookup (Match filters) - range: Enable range queries

Delete a payload field index. Args: - collectionName: Name of the collection - fieldName: Payload field to remove index from

Set or update payload fields for points. Merges with existing payload (does not replace). Args: - collectionName: Name of the collection - payload: Payload fields to set - ids: Point IDs to update (optional if using filter) - filter: Filter to select points (optional if using ids) - wait: Wait for operation to complete

Replace entire payload for points. Completely replaces existing payload (does not merge). Args: - collectionName: Name of the collection - payload: New payload to set - ids: Point IDs to update - filter: Filter to select points

Compute distance matrix in sparse matrix format. Returns distances as offset arrays for efficient storage.

Delete specific payload fields from points. Removes specified keys from payload. Args: - collectionName: Name of the collection - keys: Payload keys to delete - ids: Point IDs to update - filter: Filter to select points

Clear all payload from points. Removes entire payload, keeping only vectors. Args: - collectionName: Name of the collection - ids: Point IDs to clear - filter: Filter to select points

Upsert (insert or update) points into a collection. Points with existing IDs will be overwritten. Args: - collectionName: Target collection - points: Array of points with id, vector, and optional payload - wait: Wait for operation to complete (default: true)

Retrieve points by their IDs. Args: - collectionName: Name of the collection - ids: Array of point IDs to retrieve - withPayload: Include payload in response (default: true) - withVector: Include vectors in response (default: false)

Retrieve a single point by ID. Args: - collectionName: Name of the collection - id: Point ID to retrieve

Delete points from a collection. Can delete by IDs or by filter. Args: - collectionName: Name of the collection - ids: Array of point IDs to delete (optional) - filter: Filter to select points to delete (optional) - wait: Wait for operation to complete (default: true)

Scroll through points in a collection with pagination. Useful for iterating over all points or a filtered subset. Args: - collectionName: Name of the collection - filter: Filter to select points (optional) - limit: Number of points to return (default: 10) - offset: Point ID to start from (for pagination) - withPayload: Include payload in response (default: true) - withVector: Include vectors in response (default: false) - orderBy: Order results by payload field

Count points in a collection. Can count all points or filtered subset. Args: - collectionName: Name of the collection - filter: Filter to select points (optional) - exact: Return exact count (slower) vs estimate (default: true)

Search for similar vectors in a collection. Performs approximate nearest neighbor search. Args: - collectionName: Name of the collection - vector: Query vector (array of numbers) - limit: Number of results to return - filter: Filter to narrow search (optional) - withPayload: Include payload in results (default: true) - withVector: Include vectors in results (default: false) - scoreThreshold: Minimum score threshold (optional) - params: Search parameters (hnsw_ef, exact)

Get Qdrant server version and build information. Returns version, commit hash, and build info.

Search using a named vector. For collections with multiple named vectors. Args: - collectionName: Name of the collection - vectorName: Name of the vector field to search - vector: Query vector - limit: Number of results

Perform multiple searches in a single request. More efficient than multiple individual searches. Args: - collectionName: Name of the collection - searches: Array of search requests

Universal query endpoint supporting various search modes. Supports dense vectors, sparse vectors, and search by ID. Args: - collectionName: Name of the collection - query: Query vector or point ID - using: Named vector to use (optional) - limit: Number of results - filter: Filter conditions

Query with results grouped by a payload field. Useful for grouping results by document, category, etc. Args: - collectionName: Name of the collection - query: Query vector - groupBy: Payload field to group by - groupSize: Number of results per group - limit: Number of groups to return

Get recommendations based on positive and negative examples. Finds points similar to positive examples and dissimilar to negative ones. Args: - collectionName: Name of the collection - positive: Point IDs or vectors to use as positive examples - negative: Point IDs or vectors to use as negative examples (optional) - strategy: Recommendation strategy (average_vector or best_score) - limit: Number of results

Get recommendations grouped by a payload field.

Discover points using a target and context pairs. Uses context pairs to define regions and finds points similar to target within those regions. Args: - collectionName: Name of the collection - target: Target point ID or vector - context: Array of positive/negative pairs defining the search space - limit: Number of results

Compute distance matrix between points as pairs. Returns distances in pairwise format (point A, point B, score). Args: - collectionName: Name of the collection - sample: Sample size (optional) - limit: Maximum pairs to return - filter: Filter to select points

Get telemetry data from Qdrant. Returns information about collections, cluster state, and request statistics. Args: - anonymize: Anonymize sensitive data (default: true)

Get Prometheus-format metrics from Qdrant. Returns metrics in Prometheus/OpenMetrics format.

Get current lock status. Returns whether write operations are locked.

Set service locks. Can lock write operations for maintenance. Args: - write: Lock write operations - errorMessage: Message to return when locked operations are attempted

List all snapshots for a collection. Returns snapshot names, sizes, and creation times.

Create a snapshot of a collection. Snapshots are point-in-time backups. Args: - collectionName: Name of the collection - wait: Wait for snapshot creation to complete

Delete a collection snapshot. Args: - collectionName: Name of the collection - snapshotName: Name of the snapshot to delete

Recover a collection from a snapshot. Can recover from a URL or local file path. Args: - collectionName: Name of the collection to recover - location: Snapshot URL or file path - priority: Recovery priority (snapshot, replica, no_sync)

List all full storage snapshots. Full snapshots include all collections and aliases.

Create a full storage snapshot. Includes all collections and aliases.

Delete a full storage snapshot.

Update vectors for existing points. Replaces vectors without modifying payload. Args: - collectionName: Name of the collection - points: Array of {id, vector} pairs to update - wait: Wait for operation to complete

Delete specific named vectors from points. For collections with multiple named vectors, removes specified vectors while keeping others. Args: - collectionName: Name of the collection - vectorNames: Names of vectors to delete - ids: Point IDs to update - filter: Filter to select points