statsig-io · devin-ai-integration · Nov 15, 2025
diff --git a/README.mdx b/README.mdx
@@ -1,5 +1,12 @@
+---
+title: Readme
+description: <h1 align="center"> <a href="https://statsig.com/?ref=gh_docs"> <img width="300px" src="/images/Statsig_Favicon_Dark.png" /> </a> <div /> <a href="htt
+---
 ## Welcome to the Statsig Docs
 
+This page explains welcome to the statsig docs.
+
+
 <h1 align="center">
 	<a href="https://statsig.com/?ref=gh_docs">
 		<img width="300px" src="/images/Statsig_Favicon_Dark.png" />
@@ -8,7 +15,7 @@
 	<a href="https://statsig.com/?ref=gh_docs">Statsig</a>
 </h1>
 
-Statsig empowers you to ship, measure, and learn from your releases using the same tools as the best tech companies in the world. With Statsig, you can run thousands of A/B tests, safely rollout features, and dive deep on core business metrics and user behavior—all on a single, unified platform. To get started with Statsig, visit our [website](https://statsig.com?ref=gh_docs) or [sign up](https://console.statsig.com/sign_up?ref=gh_docs).
+Statsig empowers you to ship, measure, and learn from your releases using the same tools as the best tech companies in the world. With Statsig, you can run thousands of Experiments, safely rollout features, and dive deep on core business metrics and user behavior—all on a single, unified platform. To get started with Statsig, visit our [website](https://statsig.com?ref=gh_docs) or [sign up](https://console.statsig.com/sign_up?ref=gh_docs).
 
 This repository is hosted at https://docs.statsig.com. 
 
@@ -20,11 +27,11 @@ There are two ways to contribute:
 1. Open a Pull Request on Github (via local dev, or by editing in the Github UI)
 2. Use the Mintlify Visual Editor (Statsig Employees Only)
 
-If you're used to docs-as-code or our old docusaurus setup, #1 is probably the right path - see below for local development instructions. If you'd prefer a snazzy visual interface, message Brock on slack to get setup with a Visual Editor account. 
+If you're used to docs-as-code or our old docusaurus setup, #1 is probably the right path - refer to the following example for local development instructions. If you'd prefer a snazzy visual interface, message Brock on slack to get setup with a Visual Editor account. 
 
 ### Local development:
 
-```
+```bash
 npm i -g mintlify
 mintlify dev
 ```
@@ -47,4 +54,4 @@ You can find the allowlist of non-dictionary words in styles/config/vocabularies
 #### Troubleshooting
 
 - Mintlify dev isn't running - Run `mintlify install` it'll re-install dependencies.
-- Page loads as a 404 - Make sure you are running in a folder with `docs.json`
+- Page loads as a 404 - Make sure you are running in a folder with `docs.json`
diff --git a/access-management/discussions.mdx b/access-management/discussions.mdx
@@ -1,5 +1,6 @@
 ---
 title: Discussions in Statsig
+description: Feature rollout and experimentation are collaborative exercises where teams work together. Often this collaboration requires people taking screenshots
 ---
 
 ## Discussions

diff --git a/access-management/guide.mdx b/access-management/guide.mdx
@@ -1,5 +1,6 @@
 ---
 title: Initial Setup Guide of your Workspace
+description: <Info> Organizations and their related features are for Enterprise contracts only. Please reach out to our [support team](mailto:[email protected]),
 ---
 
 <Info>
@@ -25,7 +26,7 @@ In Statsig, we have three constructs to help you organize your workspace and sca
 
 **Organization** is an enterprise-level environment that allows companies to create project(s) and bring members to work inside the project.
 
-**Project** is a workspace within the organization where the configs (e.g., feature gates, experiments, layers, etc.), metrics, and SDK keys you and your team created lives.
+**Project** is a workspace within the organization where the configs (e.g., feature flags, experiments, layers, etc.), metrics, and SDK keys you and your team created lives.
 
 **Team** is a group of members at the project level that can help your organization manage the permissions and ownership of resources.
 
@@ -35,7 +36,7 @@ In Statsig, we have three constructs to help you organize your workspace and sca
 In Statsig, each project within the organization is isolated from one another.
 This means that **none** of the resources or data is shared among different projects, even if they are part of the same organization.
 
-The Statsig team believe that the **sensible default structure** for most customers is having an organization with a **single project** where multiple teams contribute and collaborate inside.
+The Statsig team believe that the **sensible default structure** for most users is having an organization with a **single project** where multiple teams contribute and collaborate inside.
 
 We believe that different teams and functions across your organization can stay well-organized within a single project by leveraging features such as **[teams](https://docs.statsig.com/access-management/teams), [roles](https://docs.statsig.com/access-management/projects#roles), [tags](https://docs.statsig.com/access-management/tags),** and **[templates](https://docs.statsig.com/experiments/templates/templates)**.
 This also removes the risk of costly migrations if some projects ever need to be consolidated in the future for cross-functional collaborations.
@@ -73,7 +74,7 @@ New users who are provisioned via SSO will be assigned the *Member* Role unless
   <img src="/images/access_management_project_view.png" alt="Organization project administration interface" />
 </Frame>
 
-A project in Statsig serves as a workspace that contains everything you and your team will create. This includes configs (e.g., feature gates, experiments, dynamic configs, layers), metrics, integrations, and more.
+A project in Statsig serves as a workspace that contains everything you and your team will create. This includes configs (e.g., feature flags, experiments, dynamic configs, layers), metrics, integrations, and more.
 
 When creating a new project, you can set its type to *Open* which would allow anyone in the organization to join freely, or to *Closed* which would allow people to join only by invitation or request.
 
@@ -140,7 +141,7 @@ Each **team** also has team-level **review settings** that can require reviews f
   <img src="/images/access_management_sdk_env.png" alt="SDK environment interface" />
 </Frame>
 
-In Statsig, you have **Client API Keys** to initialize all Statsig [client SDKs](https://docs.statsig.com/client/introduction) and **Server Secret Keys** to initialize all Statsig [server SDKs](https://docs.statsig.com/server/introduction).
+In Statsig, you have **Client API Keys** to initialize all Statsig [client SDKs](https://docs.statsig.com/client/introduction) and **Api Key Keys** to initialize all Statsig [server SDKs](https://docs.statsig.com/server/introduction).
 
 We believe there is a **"Crawl, Walk, Run"** phase when it comes to configuring your API keys:
 

diff --git a/access-management/introduction.mdx b/access-management/introduction.mdx
@@ -1,6 +1,7 @@
 ---
 title: Workspace Management Overview
 sidebarTitle: Overview
+description: Statsig provides a few different solutions for access management as you scale out adoption in your team/org/company. We have simple settings like auto
 ---
 
 Statsig provides a few different solutions for access management as you scale out adoption in your team/org/company.

diff --git a/access-management/org-admin/experiment_policy.mdx b/access-management/org-admin/experiment_policy.mdx
@@ -1,5 +1,6 @@
 ---
 title: Experiment Policy
+description: <Info> Organization level Experiment Policies are an Enterprise only feature. </Info> * Set defaults for new experiments: While Statsig aims to provid
 ---
 
 <Info>

diff --git a/access-management/org-admin/gates_policy.mdx b/access-management/org-admin/gates_policy.mdx
@@ -1,5 +1,6 @@
 ---
 title: Feature Gates Policy
+description: <Info> Organization level Feature Gate Policies are an Enterprise only feature. </Info> Feature Gates Policy grants organization admins the ability to
 ---
 
 <Info>

diff --git a/access-management/org-admin/organization_policies.mdx b/access-management/org-admin/organization_policies.mdx
@@ -1,5 +1,6 @@
 ---
 title: Organization Policies
+description: <Info> Organization-level Experiment and Gate Policies are an Enterprise only feature. </Info> You can configure Organization-level Experiment and Gat
 ---
 
 <Info>

diff --git a/access-management/organizations.mdx b/access-management/organizations.mdx
@@ -1,5 +1,6 @@
 ---
 title: Organization Settings & Administration
+description: <Info> Organizations and their related features are for Enterprise contracts only. Please reach out to our [support team](mailto:[email protected]),
 ---
 
 <Info>
@@ -27,7 +28,7 @@ You can view your Organization page by navigating to [Account Settings](https://
   <img src="/images/access_management_account_info.png" alt="Account Settings organization management interface" />
 </Frame>
 
-#### Organization Information
+### Organization Information
 Your organization's **Info** sidebar includes the organization's name, SSO configuration, and other settings on access management and security settings.
 
 As the organization's **Admin**, you can enable or disable SSO for all projects in the organization from this one place. 

diff --git a/access-management/projects.mdx b/access-management/projects.mdx
@@ -1,5 +1,6 @@
 ---
 title: Project Access Management
+description: <Info> This guide applies only to our on-demand customers. If you are an organization who has set up SSO, this guide will not apply to you. Please see
 ---
 <Info>
 This guide applies only to our on-demand customers. If you are an organization who has set up SSO, this guide will not apply to you. Please see our [SSO Guides](/access-management/sso/overview) for more information about how to manage access permissions through SSO.

diff --git a/access-management/scim/concepts.mdx b/access-management/scim/concepts.mdx
@@ -1,5 +1,6 @@
 ---
 title: SCIM Concepts
+description: <br /> Our SCIM implementation represents both Statsig users at the Organization level and Project level with their associated roles. There are two ma
 ---
 
 

diff --git a/access-management/scim/okta_scim_org_roles.mdx b/access-management/scim/okta_scim_org_roles.mdx
@@ -1,6 +1,7 @@
 ---
 title: Okta SCIM Org Roles
 sidebarTitle: Org Roles
+description: For every user, Statsig surfaces a SCIM field named `statsigOrgRole`. Through this field, you can manage organization user roles. Currently, Okta can 
 ---
 
 

diff --git a/access-management/scim/okta_scim_setup.mdx b/access-management/scim/okta_scim_setup.mdx
@@ -1,6 +1,7 @@
 ---
 title: Okta SCIM Setup
 sidebarTitle: Setup
+description: This guide outlines the process for setting up SCIM (System for Cross-domain Identity Management) integration between Statsig and Okta. This integrati
 ---
 This guide outlines the process for setting up SCIM (System for Cross-domain Identity
 Management) integration between Statsig and Okta. This integration allows for automated

diff --git a/access-management/scim/okta_scim_team_management.mdx b/access-management/scim/okta_scim_team_management.mdx
@@ -1,6 +1,7 @@
 ---
 title: Okta SCIM Team Management
 sidebarTitle: Team Management
+description: <Note> Teams are specific Statsig groups that exist within projects. They are not shared across projects. They also have two possible roles: `Admin` a
 ---
 
 

diff --git a/access-management/scim/okta_scim_troubleshooting.mdx b/access-management/scim/okta_scim_troubleshooting.mdx
@@ -1,6 +1,7 @@
 ---
 title: Okta SCIM Troubleshooting
 sidebarTitle: Troubleshooting
+description: | Error | Solution | |-------|----------| | User userID is not allowed to be created in this organization | Ensure that the users you are assigning to
 ---
 
 

diff --git a/access-management/scim/okta_scim_user_management.mdx b/access-management/scim/okta_scim_user_management.mdx
@@ -1,6 +1,7 @@
 ---
 title: Okta SCIM User and Project/Role Management
 sidebarTitle: User and Project/Role Management
+description: <Note> Users not assigned to the integration cannot be pushed into groups. </Note> - In Okta, go to the Statsig app's "Import" tab
 ---
 
 

diff --git a/access-management/scim/overview.mdx b/access-management/scim/overview.mdx
@@ -1,6 +1,7 @@
 ---
 title: SCIM User Provisioning
 sidebarTitle: Overview
+description: SCIM (System for Cross-domain Identity Management) is a standardized protocol that simplifies the automation of user provisioning and management acros
 ---
 
 

diff --git a/access-management/scim/scim-endpoints.mdx b/access-management/scim/scim-endpoints.mdx
@@ -1,6 +1,7 @@
 ---
 title: SCIM API Overview
 sidebarTitle: API Overview
+description: The System for Cross-domain Identity Management (SCIM) specification is designed to make managing user identities in cloud-based applications and serv
 ---
 
 The System for Cross-domain Identity Management (SCIM) specification is designed to make managing user identities in cloud-based applications and services easier. Statsig's SCIM API allows you to automate user provisioning and deprovisioning between your identity provider and Statsig.

diff --git a/access-management/sso/azuread.mdx b/access-management/sso/azuread.mdx
@@ -1,5 +1,6 @@
 ---
 title: Single Sign-On With Entra ID/Azure AD/Office 365
+description: Microsoft Entra ID, formerly known as Azure AD, is a supported IdP for SSO into Statsig. - You will need to be the `Admin` of the Statsig Project you 
 ---
 Microsoft Entra ID, formerly known as Azure AD, is a supported IdP for SSO into Statsig.
 

diff --git a/access-management/sso/google.mdx b/access-management/sso/google.mdx
@@ -1,5 +1,6 @@
 ---
 title: SSO with Google as your IdP
+description: - You will need to be the `Admin` of the Statsig Project you intend to add SSO with Google Apps to.
 ---
 ## Requirements
 

diff --git a/access-management/sso/okta_sso.mdx b/access-management/sso/okta_sso.mdx
@@ -1,5 +1,6 @@
 ---
 title: Single Sign-On With Okta
+description: - You will need to be the `Admin` of the Statsig Organization you intend to add SSO with Okta to. - You will need to be the Administrator of the Okta 
 ---
 
 ## Requirements

diff --git a/access-management/sso/overview.mdx b/access-management/sso/overview.mdx
@@ -1,5 +1,6 @@
 ---
 title: Single Sign-On With OIDC
+description: <Info> SSO is an Enterprise feature. Please reach out to our [support team](mailto:[email protected]), your sales contact, or via our slack channel 
 ---
 
 <Info>

diff --git a/access-management/tags.mdx b/access-management/tags.mdx
@@ -1,5 +1,6 @@
 ---
 title: Tags in Statsig
+description: Tags let you apply light-weight organization to your Statsig config (e.g. gates, experiments and metrics) to allow easy filtering by team (or organiza
 ---
 
 ## Tags for organization

diff --git a/access-management/teams.mdx b/access-management/teams.mdx
@@ -1,5 +1,6 @@
 ---
 title: Teams
+description: <Info> Teams are an Enterprise-only feature. If you are on the Developer or Pro tiers, this guide will not apply to you. To upgrade to Enterprise, fee
 ---
 <Info>
 Teams are an Enterprise-only feature. If you are on the Developer or Pro tiers, this guide will not apply to you. To upgrade to Enterprise, feel free to reach out to our team [here](https://www.statsig.com/contact/demo).

diff --git a/ai-evals/offline-evals.mdx b/ai-evals/offline-evals.mdx
@@ -1,5 +1,6 @@
 ---
 title: Offline Evals
+description: Offline evals offer a quick, automated grading of model outputs on a fixed test set. They catch wins / regressions early—before any real users are exp
 ---
 
 ## What are Offline Evals

diff --git a/ai-evals/online-evals.mdx b/ai-evals/online-evals.mdx
@@ -1,5 +1,6 @@
 ---
 title: Online Evals
+description: Online evals let you grade your model output in production on real world use cases. You can run the "live" version of a prompt, but can also shadow ru
 ---
 
 ## What are Online Evals
@@ -59,7 +60,7 @@ const liveOutput = client.completions.create(
 );
 
 // simulateneously run completions on the candidate prompts to get their output
-```
+```kotlin
 
 **3. Score your output using graders**
 

diff --git a/ai-evals/overview.mdx b/ai-evals/overview.mdx
@@ -1,6 +1,7 @@
 ---
 title: AI Evals Overview
 sidebarTitle: Overview
+description: <Info> AI Evals are currently in beta; Reach out in Slack to get access. </Info> Statsig AI Evals have a few core components to help iterate and serve
 ---
 
 <Info>

diff --git a/ai-evals/prompts.mdx b/ai-evals/prompts.mdx
@@ -1,6 +1,7 @@
 ---
 title: Prompts & Graders
 sidebarTitle: Prompts & Graders
+description: A Prompt is a way to represent an LLM prompt or a task in Statsig, with it's config. Prompts are similar to Dynamic Configs, and allow you to evaluate
 ---
 
 ## What is a Prompt in Statsig?

diff --git a/autotune/bandit-faq.mdx b/autotune/bandit-faq.mdx
@@ -5,6 +5,7 @@ keywords:
   - owner:craig
 last_update:
   date: 2025-09-18
+description: You will see diagnostic data appear in near-real time in the logstream on your bandit. Data on the models/results section will depend on your settings
 ---
 
 ### When should I see data show up in a bandit?

diff --git a/autotune/bandit-introduction.mdx b/autotune/bandit-introduction.mdx
@@ -5,6 +5,7 @@ keywords:
   - owner:vm
 last_update:
   date: 2025-09-18
+description: Multi-Armed Bandits are solutions that automatically find the best variant among a group of candidates, balancing between "exploring" options and "exp
 ---
 
 Multi-Armed Bandits are solutions that automatically find the best variant among a group of candidates, balancing between "exploring" options and "exploiting" the best option by dynamically allocating traffic. On Statsig, Bandits are used to pick the best user experience to drive a target metric or action.

diff --git a/autotune/contextual-bandit.mdx b/autotune/contextual-bandit.mdx
@@ -5,6 +5,7 @@ keywords:
   - owner:vm
 last_update:
   date: 2025-09-18
+description: Autotune UI implements a variant of the LinUCB algorithm. This estimates a user's outcome for each variant, and incorporates the model's uncertainty t
 ---
 
 ## Model

diff --git a/autotune/contextual/getting-started.mdx b/autotune/contextual/getting-started.mdx
@@ -5,6 +5,7 @@ keywords:
   - owner:vm
 last_update:
   date: 2025-09-18
+description: Getting started with Autotune AI can be done very quickly. <Warning> Statsig supports contextual autotune in all Client SDKs, but only in the followin
 ---
 
 Getting started with Autotune AI can be done very quickly.
@@ -99,7 +100,7 @@ We assume you have your server secret key for the following code. Before running
 
 First, import and initialize Statsig:
 
-```
+```javascript
 from statsig_python_core import Statsig, StatsigUser
 
 key = <your_key_here>
@@ -111,14 +112,14 @@ statsig.initialize().wait()
 
 Then, create a user object and fetch your config:
 
-```
+```text
 user = StatsigUser('user_id', custom={'key1': 'value1', 'key2': 'value2'})
 cfg = statsig.get_experiment(user, autotune_name)
 ```
 
 Now you have your cfg and can apply it!
 
-```
+```python
 color = cfg.get_string("color", "default color")
 print(f"Going to use {color} for my color now")
 ```
@@ -134,7 +135,7 @@ That's it! Your code is now serving personalized variants to your users.
 
 Statsig requires a few hundred units to train a model, and will also not start training until those units' attribution window has elapsed. If you want to test the functionality, we highly recommend "faking a test" to confirm things work like you expect - use logic like
 
-```
+```text
 fetch_autotune_value()
 if(user country == 'us'):
     log_click()

diff --git a/autotune/contextual/introduction.mdx b/autotune/contextual/introduction.mdx
@@ -5,6 +5,7 @@ keywords:
   - owner:vm
 last_update:
   date: 2025-09-18
+description: Contextual Multi-Armed Bandits are a subset of Multi-Armed-Bandits which use context about a user to personalize their experience. This is generally a
 ---
 
 Contextual Multi-Armed Bandits are a subset of Multi-Armed-Bandits which use context about a user to personalize their experience. This is generally achieved by predicting outcomes from among the variants, and picking the best outcome while factoring for uncertainty. Specifically, they will tend to prefer a slightly worse prediction that has a lot of uncertainty, thereby exploring that variant.

diff --git a/autotune/contextual/methodology.mdx b/autotune/contextual/methodology.mdx
@@ -5,6 +5,7 @@ keywords:
   - owner:craig
 last_update:
   date: 2025-09-18
+description: This page covers the high level approach that Statsig takes to running contextual bandits across cloud and warehouse native. Specifics of implementati
 ---
 
 ## Methodology

diff --git a/autotune/contextual/monitoring.mdx b/autotune/contextual/monitoring.mdx
@@ -5,6 +5,7 @@ keywords:
   - owner:vm
 last_update:
   date: 2025-09-18
+description: There are three primary ways we recommend you monitor autotune performance. The best way to evaluate if a bandit is working is seeing if it drives mor
 ---
 
 There are three primary ways we recommend you monitor autotune performance.
@@ -17,7 +18,7 @@ This is the gold standard of measurement and is highly encouraged.
 
 Standard practice is to wrap the autotune in a experiment with a binary parameter, either as 50/50 or a 90/10 holdback.  You can link the experiment to the autotune to get the results on the autotune page. In code, this might look like:
 
-```
+```yaml
 experiment_value = statsig.get_experiment('wrapping_experiment').get('flag')
 default_param = '..."
 if(experiment_value):