Statsig MCP

Statsig is a feature flagging, experimentation, and product analytics platform, and its MCP server exposes that control plane to an AI client. Rather than switching to the console to work out which gate governs a code path, a developer or an agent queries the configuration from an editor or chat session and gets back the rule set, the targeting conditions, and which environments a change is live in. The server covers reads and writes. On the read side it surfaces feature gates, experiments, dynamic configs, segments, and the metrics attached to them, including how an experiment is trending against its primary metric. On the write side it can create gates, adjust rollout percentages, and modify targeting rules. That combination is what makes it useful during an incident: the agent locates the flag behind a failing code path and dials it back without a human hunting through the dashboard first. The natural users are platform and product engineering teams that already ship behind flags, plus data teams reading experiment results. Because one server can both read results and change rollout state, most teams scope the API key so agents read broadly but write only to the environments they are meant to touch.