Statsig MCP
Statsig MCP exposes experiments, feature gates, dynamic configs, layers, metrics, segments, reviews, audit logs, and observability workflows.
Verdict
Common use cases
- Check A/B test results during sprint planning
- Verify feature flag rollout percentages
- Compare metric trends across experiment arms
- Audit gate overrides in staging environments
- Summarize experiment performance for stakeholders
Integration
- Vendor
- Statsig MCP
- Category
- other
- Auth
- OAUTH2
- Composio slug
statsig_mcp
Tools
Setup
Setup guide
- 11. Inside Switchy, open your workspace settings and navigate to the Integrations tab. 2. Find Statsig MCP in the list and click Connect. 3. You'll redirect to Statsig's OAuth consent screen — log in with an account that has at least read permissions on the projects you want to query. 4. Grant the requested scopes (typically read access to experiments, gates, and metrics). 5. Statsig redirects you back to Switchy, where you'll see a green Connected badge. 6. Open any Project and type '@Statsig what's the conversion rate for the checkout redesign experiment?' to test the connection. 7. If the AI returns current data, you're set — if it errors, check that your Statsig account has access to the project you're querying.
What teammates see: by default, memories from Statsig MCP are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Experiment Status Check
@Statsig show me the latest results for the new onboarding flow experiment, including conversion rate and confidence intervals for each variant.
Feature Gate Audit
@Statsig list all feature gates currently enabled in production, showing rollout percentage and target user segments for each.
Metric Comparison
@Statsig compare average session duration between control and treatment groups in the homepage redesign test over the past 7 days.
Rollout Timeline
@Statsig show me the rollout timeline for the new checkout feature gate, including when percentages changed and who made the updates.
Experiment Summary
@Statsig summarize the payment flow experiment results in simple terms, highlighting which variant performed best and whether we should ship it.
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Statsig pull the last 7 days of metrics for the 'premium_upsell_banner' experiment and summarize which variant is winning on conversion rate
The 'premium_upsell_banner' experiment has three variants over 8,420 users. Variant B (larger CTA button) leads with 4.2% conversion rate, +31% vs. control (3.2%). Variant C (testimonial copy) shows 3.8%, +19% vs. control. Statistical significance: B is significant at p<0.05, C is not yet significant. Recommendation: consider ramping Variant B to 50% if business context supports the larger button design.
This synthesis example pairs Statsig's experiment data with AI reasoning to surface a recommendation. The AI interprets statistical significance but cannot account for your product strategy — use this as a starting point for discussion, not a final decision. Requires read access to experiment results.
Use-case deep-dives
When Statsig MCP helps product teams ship safer releases
A 5-person product team ships weekly and uses feature flags to gate new UI changes. The Statsig MCP lets them check flag states, toggle rollouts, and pull experiment metrics without leaving their AI workspace during standups or post-mortems. This works when your team already runs Statsig and needs quick flag lookups or rollout adjustments in conversation threads. The OAuth2 flow means each person authenticates once and the AI can act on their behalf. The trade-off: if you're just reading metrics dashboards or doing deep statistical analysis, you're better off in Statsig's web console where the visualizations are richer. Use this MCP when your team's workflow is chat-first and you need to make flag decisions in the same thread where you're discussing the release.
When this MCP speeds up A/B test review cycles
A growth team runs 8-12 concurrent A/B tests and reviews results every Monday in a shared chat workspace. The Statsig MCP pulls experiment metrics, checks statistical significance, and surfaces winning variants without context-switching to the Statsig dashboard. This is the right call when your team's decision rhythm is synchronous and chat-native—everyone sees the same data in the same thread and can react immediately. The MCP handles the lookup and formatting so the conversation stays in one place. The boundary: if your experiments involve complex segmentation or you need to drill into user-level data, the web UI is still faster. Use this MCP when the bottleneck is switching tools mid-discussion, not the depth of analysis.
When Statsig MCP helps support teams debug user-specific issues
A 3-person support team fields tickets where users report inconsistent behavior—often caused by feature flags targeting specific cohorts. The Statsig MCP lets them check which flags are active for a given user ID or email without pinging engineering or opening the Statsig console. This works when your support team has read access to Statsig and the issue is 'why is this user seeing X when others see Y'. The OAuth2 setup means each support agent authenticates with their own Statsig permissions. The limit: if the issue requires changing flag targeting rules or rolling back a release, that still needs engineering or product to act in the full Statsig UI. Use this MCP when your support workflow is triage-first and you need to answer 'what flags is this user in' in under 60 seconds.
Frequently asked
What does the Statsig MCP do in Switchy?
The Statsig MCP connects your team's Switchy workspace to Statsig's feature flag and experimentation platform. Your AI agents can query experiment results, check feature gate status, and pull metrics without leaving the conversation. This means you can ask questions about A/B test performance or flag rollout status and get answers directly in Switchy instead of switching to the Statsig dashboard.
Does Statsig MCP require OAuth or can I use an API key?
Statsig MCP uses OAuth2, so you'll authorize through Statsig's login flow rather than copying API keys. The OAuth connection means Switchy inherits your existing Statsig permissions — if you can't view an experiment in Statsig's UI, the MCP won't surface it either. You'll need a Statsig account with at least read access to the projects you want to query.
Can the Statsig MCP create or modify feature flags?
The current Statsig MCP implementation focuses on read operations — querying experiment results, checking gate status, pulling metric snapshots. If you need to create flags or change targeting rules, you'll still use Statsig's dashboard or their REST API directly. This read-only approach keeps your production feature flags safe from accidental AI edits during exploratory conversations.
How is this different from just opening Statsig in another tab?
The MCP lets you ask natural-language questions about your experiments without context-switching. Instead of navigating Statsig's UI to compare conversion rates across three variants, you ask Switchy and get a direct answer. It's faster for ad-hoc questions during team discussions, though you'll still want the full Statsig dashboard for detailed analysis or configuration changes.
Who on my team should connect the Statsig MCP?
Anyone who regularly checks experiment results or feature flag status — typically product managers, engineers, or data analysts. The person connecting it should have Statsig access to the projects your team cares about, since the MCP inherits their permissions. Multiple team members can connect their own Statsig accounts if you want different permission scopes across your workspace.