Provider REST API

15 actions

Connect Claude to Qdrant

Search, read and write points in Qdrant vector collections, and manage the collections and their payload indexes. Toolspoke puts 15 of its actions behind one MCP endpoint that Claude, Cursor, and Codex all speak.

Connect View pricing

Connection: Provider REST API
Authentication: API key
Actions exposed: 15
What a call bills: Nothing
Adapter: Maintained by Toolspoke

Connected in three steps

1
Install Qdrant
Open the marketplace in your workspace, add Qdrant to the project your agents work in, and it appears on the gateway immediately.
2
Connect the credential
Authenticate with api key. Where to get one, and what it has to be able to reach, is the next section.
3
Point your agent at the gateway
Give your client one address, https://toolspoke.com/mcp. Claude Code takes it as a command, Claude and Claude Desktop add it as a custom connector, and Cursor, Codex and VS Code each read it from a config file of their own.

.mcp.json

{
  "mcpServers": {
    "toolspoke": {
      "type": "http",
      "url": "https://toolspoke.com/mcp"
    }
  }
}

One block covers every tool you have installed. Qdrant shows up in the client as soon as your policy allows it, and so does everything else you install later.

Where the address goes, per client

Claude Code: Run it in your project, then /mcp to sign in
claude mcp add --transport http toolspoke https://toolspoke.com/mcp
Claude and Claude Desktop: Settings, then Connectors, then Add custom connector
https://toolspoke.com/mcp
Cursor: ~/.cursor/mcp.json, or .cursor/mcp.json for one project
{ "mcpServers": { "toolspoke": { "url": "https://toolspoke.com/mcp" } } }
Codex: ~/.codex/config.toml
[mcp_servers.toolspoke] url = "https://toolspoke.com/mcp"
VS Code: .vscode/mcp.json, or the MCP: Add Server command
{ "servers": { "toolspoke": { "type": "http", "url": "https://toolspoke.com/mcp" } } }

What Qdrant asks for

API key. You provide it once, when you install the connector. Toolspoke encrypts it at rest and decrypts it only for the length of a single call, and the gateway attaches it to the outbound request itself, so it is never part of the arguments an agent sends.

Cluster URLRequired: Qdrant Cloud → Clusters → your cluster → the Endpoint shown on its overview, with :6333 at the end. A self-hosted server is usually http://host:6333 and must be reachable from the internet. No path.; https://xyz-example.eu-central.aws.cloud.qdrant.io:6333
API keyRequired: Qdrant Cloud → Clusters → your cluster → API Keys → Create. A read-only key is enough for every read here; writes need a read-write key. Self-hosted: the service.api_key (or read_only_api_key) from the server's config.; eyJhbGciOi…

What Claude can do in Qdrant

15 actions, each one declared and named by the connector rather than discovered at runtime. A workspace policy grants a person all of them, a hand-picked selection, everything on the read side, everything on the write side, or none.

Reads: 9Reads
Writes: 4Writes
Destructive: 2Destructive

Reads

Fetches data and changes nothing.

list_collections
List the names of every collection on the cluster. Call this first to confirm the connection and find what to query.
get_collection
Read one collection's status, point count, vector configuration (size, distance, named vectors), payload indexes and optimizer state. Check the vector names and sizes here before querying.
collection_exists
Check whether a collection exists, without reading its configuration.
list_aliases
List every collection alias on the cluster and the collection each one points to.
query_points
Search a collection with Qdrant's universal query: nearest neighbours to a vector, recommendation from point ids, or an ordered or filtered listing. Pass `query` as a vector (a list of floats) for similarity search, or as a point id to search near that point. Qdrant does not embed text here - the vector must come from the same model the collection was built with.
scroll_points
Page through a collection's points in id order, optionally filtered by payload. Returns next_page_offset; pass it back as `offset` for the next page.
count_points
Count the points in a collection, or the ones matching a payload filter.
get_points
Read specific points by id, with their payloads.
facet
Count how many points have each value of one payload key, like a GROUP BY. The key needs a keyword or integer payload index.

Writes

Creates or updates something on the other side.

create_collection
Create a collection. `vectors` sets the vector size and distance, either one unnamed vector or a map of named ones. Fails if the name is taken.
create_payload_index
Index one payload key so filters and facets on it are fast. Needed for facet and for order_by in scroll_points.
upsert_points
Insert points, or replace points with the same id. Each point needs an id and a vector of the collection's size; the payload is optional JSON. Waits until the write is applied.
set_payload
Add or overwrite payload keys on points chosen by id or by filter. Keys not named are left as they are.

Destructive

Deletes or permanently alters something. Worth granting on purpose.

delete_points
Delete points by id, or every point matching a filter. There is no undo. Give exactly one of points or filter.
delete_collection
Delete a collection and every point in it. There is no undo short of restoring a snapshot.

What it will not do

Enforced by the gateway rather than left to convention, which is why each of these can be stated flatly.

It cannot call anything else: The 15 actions above are the whole of it. A call to any other name is refused before it reaches Qdrant rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time.
It reaches no further than your credential: Toolspoke holds no access to Qdrant of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either.
It never hears from Qdrant: Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Qdrant. An agent has to ask.
It does not smooth over provider limits: Toolspoke does not retry, queue or back off around Qdrant's own rate limits. A call that Qdrant refuses comes back to the agent as a failed call.

Before you connect it

What can Claude do in Qdrant?

15 named actions: 9 that only read, 4 that write and 2 that delete or permanently alter something. They include list_collections, get_collection and collection_exists. Nothing outside that list is reachable: the connector declares each operation by name rather than proxying whatever an agent asks for.

What credentials does the Qdrant connector need?

API key. The connector asks for cluster url and api key. Values are encrypted at rest and attached to the outbound request by the gateway, so they are never part of the arguments an agent sends and never reach the audit log.

Does the Qdrant connector work with Cursor and Codex, or only Claude?

Any client that speaks MCP, and every one of them gets the same 15 actions. There is a single address, https://toolspoke.com/mcp. Claude Code adds it with claude mcp add --transport http, Claude and Claude Desktop take it as a custom connector in settings, Cursor reads it from .cursor/mcp.json, Codex from ~/.codex/config.toml, and VS Code from .vscode/mcp.json. Each of them signs in to the gateway itself, so there is no key to paste.

What does the Qdrant connector not do?

The 15 actions above are the whole of it. A call to any other name is refused before it reaches Qdrant rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time. Toolspoke holds no access to Qdrant of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either. Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Qdrant. An agent has to ask. Toolspoke does not retry, queue or back off around Qdrant's own rate limits. A call that Qdrant refuses comes back to the agent as a failed call.

Can I limit which actions an agent can call?

Yes, in two places. The project switches Qdrant's actions on and off one at a time, for everyone in the project at once, and the screen groups them by read, write and destructive so turning off everything that deletes is one click. An individual agent key can then be narrowed further, to particular toolkits in a project and to particular actions in a toolkit. Whatever it was granted, a key never reaches a project its owner cannot.

What gets recorded when an agent calls Qdrant?

Every attempt, with the agent that made it and the person that agent belongs to, the full request payload, the response payload, the status, the duration, and what the call cost in sandbox time and model tokens. Values whose key names a secret are masked out before the record is shown to anyone. An operation the connector marks as not retained never has its response body written at all, so the gateway keeps no second copy of what was read.

More database connectors

Browse all of them