Provider REST API

14 actions

Connect Claude to GrowthBook

Read and toggle feature flags, and read experiments and their results, in GrowthBook. Toolspoke puts 14 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: 14
What a call bills: Nothing
Adapter: Maintained by Toolspoke

Connected in three steps

1
Install GrowthBook
Open the marketplace in your workspace, add GrowthBook 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. GrowthBook 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 GrowthBook 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.

API keyRequired: GrowthBook → your avatar → Personal Access Tokens → Create token, which acts with your own access. An admin can instead create an organization key under Settings → API Keys and give it a role - a read-only role keeps this connector to reads.; secret_… or a personal access token
API hostOptional: Leave blank for GrowthBook Cloud. For a self-hosted install, the API host (the API_HOST setting, port 3100 by default), not the app's address, and with no trailing /api - this connector adds that itself.; https://api.growthbook.io

What Claude can do in GrowthBook

14 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: 10Reads
Writes: 3Writes
Destructive: 1Destructive

Reads

Fetches data and changes nothing.

list_features
List feature flags, optionally in one project. Start here - the flag id is what every other feature action takes. It is also the install's connection test, because it needs a key that can read real data.
get_feature
Read one feature flag in full - its default value, and per environment whether it is on and which rules it serves.
list_experiments
List experiments, optionally narrowed to a project or a status.
get_experiment
Read one experiment - its hypothesis, variations, phases, metrics and status.
get_experiment_results
Read the latest analysis of one experiment - each variation's metric values and the statistics comparing them. Optionally for one phase or broken down by a dimension.
list_projects
List the projects in this organization. A project id narrows feature and experiment lists.
list_environments
List the environments in this organization, such as dev and production. Environment ids are the keys toggle_feature takes.
list_metrics
List the metrics experiments can be measured by, optionally in one project.
list_saved_groups
List saved groups - named lists of user attributes, such as user ids or emails, that targeting rules refer to. Not written to the audit log, because the lists hold other people's identifiers.
list_sdk_connections
List the SDK connections - which app reads which environment and project, and with which client key.

Writes

Creates or updates something on the other side.

create_feature
Create a feature flag. It serves its default value to everyone until rules are added, so a new flag changes no running app unless the default differs from what the code already assumes.
update_feature
Change a feature flag's description, owner, project, tags or default value. Fields not sent are left alone.
toggle_feature
Turn a flag on or off in one or more environments, e.g. environments {"production": true}. This changes what every SDK reading those environments serves at once. A write, not destructive, because sending the opposite value reverses it.

Destructive

Deletes or permanently alters something. Worth granting on purpose.

delete_feature
Delete a feature flag in every environment. Destructive because there is no undo and SDKs still reading it get no value from then on. Prefer update_feature with archived true to retire a flag reversibly.

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 14 actions above are the whole of it. A call to any other name is refused before it reaches GrowthBook 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 GrowthBook 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 GrowthBook: 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 GrowthBook. An agent has to ask.
It does not smooth over provider limits: Toolspoke does not retry, queue or back off around GrowthBook's own rate limits. A call that GrowthBook refuses comes back to the agent as a failed call.

Before you connect it

What can Claude do in GrowthBook?

14 named actions: 10 that only read, 3 that write and 1 that delete or permanently alter something. They include list_features, get_feature and create_feature. 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 GrowthBook connector need?

API key. The connector asks for api key, and optionally api host. 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 GrowthBook connector work with Cursor and Codex, or only Claude?

Any client that speaks MCP, and every one of them gets the same 14 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 GrowthBook connector not do?

The 14 actions above are the whole of it. A call to any other name is refused before it reaches GrowthBook 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 GrowthBook 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 GrowthBook. An agent has to ask. Toolspoke does not retry, queue or back off around GrowthBook's own rate limits. A call that GrowthBook 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 GrowthBook'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 GrowthBook?

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 developer tools connectors

Browse all of them