Integration · Automate and collaborate

Statsig and Meteor AI agents

Connect Statsig to an internal Met: check feature gates, experiment results, metrics and change history with Statsig's official MCP server.

Book a demo See every integration

How does Statsig connect to Meteor?

With the Statsig ready connector, your Met uses Statsig's official MCP server with the account the admin authorizes, and it gets 18 tools. It checks the connection's project and permissions, searches feature gates, experiments and dynamic configs, brings an experiment's results and reads the audit log to find out who changed what. With write access it also creates and updates gates, experiments and configs; Statsig requires every update to carry a confirmation with the reason for the change. It lets the owner or the agency ask on WhatsApp how a test is going without opening the console.

What it automates

What a Meteor Met does with Statsig

Tell you how an experiment is going

You ask the Met on WhatsApp about the new checkout test. The Met finds it, brings the overall results between the control and test groups, and sums up which metric moved and in which direction.

Find out who changed a gate

If something stopped showing in the app, the Met opens the gate, checks its version history and the project's audit log, and tells you what changed and who did it.

Turn a gate on or off if the admin allows it

With write access and the update action left on in Connections, the Met reads the whole gate and replaces it with the requested change, keeping its rules. Statsig requires that call to carry a confirmation with the reason.

Find old gates

The Met lists the project's gates, filtered by type, tag or creator, and shows you the list so your team decides which ones to retire before cleaning up code.

18 tools

Everything your Met can ask Statsig for

Each one is a real action in the connector. You can switch off the ones you would rather it never used.

Tool	What it does
`get_context`	Returns the connection's project, permissions, review settings and product context.
`search`	Searches Statsig resources with a text query, with up to 100 results.
`fetch`	Fetches the full resource picked from a search result.
`gate_read`	Lists and opens feature gates, with a rule's results and the version history.
`experiment_read`	Lists and opens experiments, with overall results, results by metric and dimension, summary charts and history.
`dynamic_config_read`	Lists and opens dynamic configs with their version history.
`metric_read`	Lists metrics and metric sources, and returns a metric's definition.
`log_read`	Queries Logs Explorer and the audit log with the project's change history.
`gate_create`	Creates a new feature gate. Needs write access.
`experiment_create`	Creates a new experiment with its name, groups and metrics. Needs write access.
`dynamic_config_create`	Creates a new dynamic config. Needs write access.
`gate_update`	Replaces a whole feature gate, including whether it is enabled and its rules. Needs write access and a confirmation.
`experiment_update`	Replaces a whole experiment, including its status. Needs write access and a confirmation.
`dynamic_config_update`	Replaces a whole dynamic config, including whether it is enabled and its rules. Needs write access and a confirmation.
`discover_tools`	Finds other Statsig operations, such as layers, segments, holdouts, tags or reviews, and returns their contract. It is read-only.
`api_read`	Runs, read-only, an operation found with discover_tools.
`api_write`	Runs a write operation found with discover_tools. Its schema does not require a confirmation. Starts switched off.
`api_destructive`	Runs a sensitive operation found with discover_tools, such as delicate updates or review actions. Requires a confirmation. Starts switched off.

The limits

What the Statsig integration does NOT do

We would rather say it here than have you find out during rollout.

It does not ask you for confirmation in the chat before writing. Statsig's confirmation is a field in the call, not a question for you: the control is which actions the admin leaves on in Connections.
It does not edit part of a resource with the update tools. They replace the whole gate, experiment or config, so the Met reads the current state first; a partial change goes through discover_tools and api_write or api_destructive, which start switched off.
It does not skip the project's permissions or reviews. If your project requires review, approving it does not apply the change by itself, and a read-only user cannot write.
It does not pay your Statsig plan. Usage runs on your Statsig account; Meteor does not resell the service.

Before you start

What you need to connect Statsig

1

A Statsig account with Personal Console API Keys enabled

Statsig's OAuth connection only works with Personal Console API Keys. The organization owner enables them for your role in the organization settings.
2

Connect it in Met

In Met, open Connections → Statsig → Connect. A Statsig window opens so you sign in and authorize Met. Then you choose which actions stay on.

Frequently asked questions

Questions about Statsig and Meteor

Yes, if the account has write access and the admin leaves the gate update action on in Connections. The tool replaces the whole gate and Statsig requires a confirmation with the reason inside the call.

You can still connect it. Statsig lets a read-only user use every read tool; the create and update tools need write permissions.

They run any operation the Met finds with discover_tools, such as changes to segments or reviews. That is why they start switched off until the admin turns them on in Connections.

You do, with your own Statsig plan. Meteor does not resell Statsig: the connector uses your company's account.

Step by step