---
title: "How to Configure Platform Settings"
description: "Configure global platform settings that control how Corridor behaves across your organization."
---

<div style="padding: 14px 16px; border: 0.5px solid #AFA9EC; border-radius: 12px; background: #EEEDFE; margin-bottom: 1.5rem;">
  <div style="display: flex; align-items: center; gap: 7px; margin-bottom: 8px;">
    <svg width="16" height="16" viewBox="0 0 16 16" fill="none">
      <path d="M9 2L4 9h4l-1 5 5-7H8l1-5z" stroke="#7f77dd" stroke-width="1.2" stroke-linecap="round" stroke-linejoin="round"/>
    </svg>
    <span style="font-size: 11px; font-weight: 500; letter-spacing: 0.05em; text-transform: uppercase; color: #534AB7;">TL;DR</span>
  </div>
  <div style="font-size: 13px; color: #3C3489; line-height: 1.65;">
    <strong>Settings → Platform Settings</strong> is the single place for instance-wide configuration: upload limits, login and SSO methods, branding, data source connectors, governance rules, default reports, and the AI assistant. Pick a section on the left, click <strong>Edit</strong>, make your change, then <strong>Save</strong>.
  </div>
</div>

:::note[Admin guide]
This page is for administrators with write access to Platform Settings. There's one settings record per instance, and changes here affect every user. Each section on the left is independent, so edit and save just the one you need.
:::

## General

Controls uploads and script validation.

* **Maximum File Size** - the largest file, in MB, a user can upload.
* **Allowed File Extensions** - a comma-separated list with leading dots (for example, `.csv, .xlsx, .pdf`). Only these types can be uploaded.
* **Allowed Python Imports** - the modules users may import in object definitions and scripts. Checked when the platform validates definition syntax.
* **Allow file upload as Input Source in Table Registry** - toggle. Lets users register tables directly from an uploaded file.
* **Enable UI Telemetry (OpenTelemetry)** - toggle. Turns on UI tracing. To send traces and metrics to Grafana, set `TELEMETRY_OTLP_ENDPOINT` in corridor-app settings and restart the server.

## Configure Logins

Sets how users sign in.

* **Allowed Login Options** - multi-select of **Password**, **Email OTP**, **Enterprise SSO**, **Google**, and **Microsoft**. At least one is required.
* **Whitelisted Domains** - comma-separated email domains allowed to sign in through a federated method (Google, Microsoft, or Enterprise SSO). Leave blank to allow any domain.
* **Default Role** - the [role](/user-guide/admin-guide/configure-roles/) assigned to a user provisioned on their first SSO login.

Whitelisted Domains and Default Role appear only once a federated login is enabled.

:::note
Changes to **Allowed Login Options** and **Whitelisted Domains** are copied to every workspace on save, so the login page reflects the enabled methods immediately.
:::

## Branding

All fields optional.

* **Financial Institution Type** - **Credit Union** or **Bank**.
* **Brand Name** - the name shown in the product.
* **Primary Logo** - logo for light backgrounds (SVG or PNG).
* **Reverse Logo** - logo for dark backgrounds (SVG or PNG).

## Hooks

* **Data Lake - Pre Read Hook** - Python that runs before the platform reads from a Data Lake source, for example to inject connection options like Snowflake credentials. The hook can use the `location` and `username` variables. Leave blank for no hook.

## Integrations

Spark connector options for each data source type: **Snowflake**, **MSSQL**, **Oracle**, **Hive**, and **Other JDBC**. Click a source's gear icon to set its options. A configured source shows an **Active** badge.

## Governance

* **Require Supporting Analysis for approvals and reviews** - toggle. When on, every approval and review must have a job attached as supporting analysis.
* **Roles with Retire Permission** - the [roles](/user-guide/admin-guide/configure-roles/) allowed to use Retire on workflowable objects.

:::caution
Turning on the supporting-analysis requirement cancels any open review requests that have no job attached. You're asked to confirm before saving.
:::

## Default Reports

For each simulatable object type, pick the reports selected by default when a simulation runs on it. Object types: **Quality Profile**, **Data Element**, **Feature**, **Model**, **Dataset**, and **Policy**.

## Single Record Review Format

Define display formats keyed by **Alias**. Each row pairs a unique **Alias** with a **Format** (`0`, `0.00`, `0.000`, `0.0000`, `0%`, or `0.00%`), which controls how that value renders in Single Record Review output.

## AI

Two sub-tabs configure the platform assistant.

### LLM Config

Sets the assistant's provider and model.

* **Provider** - **Azure OpenAI**, **OpenAI**, **Anthropic**, **Google**, **AWS Bedrock**, or **Google Vertex AI**.
* **AI Bot Name** - display name for the assistant (for example, CORRA).
* **Authentication** - fields change with the provider: an API key for most, AWS region and access keys for Bedrock, GCP project and location for Vertex AI, plus endpoint and API version for Azure OpenAI. Saved keys are masked; use **Change key** to replace one.
* **Model Name** - the model to use (labeled **Deployment Name** for Azure).
* **Temperature** - 0 to 2 (default 0.7).
* **Max Tokens** - 1 to 128000 (default 4096).
* **Enable streaming responses** - toggle, off by default. Streams the assistant's reply word by word as it generates, instead of showing it all at once when finished.

### Observability

Wires up Langfuse tracing.

* **Enable Langfuse Tracing** - toggle.
* **Public Key** - your Langfuse public key (`pk-...`).
* **Secret Key** - your Langfuse secret key (`sk-...`); masked once saved, use **Change key** to rotate.
* **Host** - optional; defaults to `https://cloud.langfuse.com`.

## What's next

* [Manage Users](/user-guide/admin-guide/manage-users/) and [Configure Roles](/user-guide/admin-guide/configure-roles/), which the login and default-role settings depend on.
* [Configure an Approval Workflow](/user-guide/admin-guide/configure-approval-workflow/), which the Governance settings interact with.