# yaml-language-server: $schema=../../../../schema/control.schema.json
id: CB-M365-POWER-007
title: Enforce Content Security Policy for model-driven and canvas apps
finding:
  title: Content Security Policy is not enforced for Power Apps
  description: Model-driven or canvas apps are served without an enforced Content Security Policy, so they can be framed by any site and get no browser-side protection against injected content.
status: active
profiles: [baseline]
risk:
  severity: medium
assessment: manual
tags: [power-platform, power-apps, csp, clickjacking, dataverse]

description: |
  In every Power Platform environment with a Dataverse database, the **Privacy + Security**
  settings enforce Content Security Policy for both app types: **Enforce content security policy**
  is **On** for **Model-driven** and for **Canvas**. The `frame-ancestors` directive keeps its
  default value (`'self' https://*.powerapps.com`) or lists only the approved hosts that embed the
  apps (Teams, Outlook, Power BI, intranet).
rationale: |
  An enforced CSP header tells the browser which origins may frame the app and load resources.
  Without it, a malicious site can embed a Power App in a hidden frame and trick users into
  clicking (clickjacking), and injected content is not constrained by the browser. Turning on
  enforcement is a low-cost, platform-provided layer of defense.
impact: |
  Apps embedded in hosts not listed in `frame-ancestors` stop loading in those hosts. Strict mode
  can block custom scripts or resources loaded from other origins. Enable reporting with an
  endpoint first (report-only mode), enforce on a test environment, then in production once no
  violation is reported. Environments without Dataverse are out of scope.

audit:
  manual: |
    1. Sign in to the Power Platform admin center (https://admin.powerplatform.microsoft.com) as at
       least a Global Reader, or as an environment admin of each environment.
    2. Browse to **Manage** > **Environments** and, for each environment with Dataverse, select it,
       then **Settings** > **Product** > **Privacy + Security**.
    3. Under **Content security policy**, check on both the **Model-driven** and **Canvas** pivots
       that **Enforce content security policy** is **On**, and review the `frame-ancestors` sources.
    4. Alternative, read only: Dataverse Web API
       `GET https://<org>.crm.dynamics.com/api/data/v9.2/organizations?$select=iscontentsecuritypolicyenabled,iscontentsecuritypolicyenabledforcanvas,contentsecuritypolicyconfiguration`
       with a user who can read the organization table.
    5. Fail if enforcement is off for either app type in any Dataverse environment, or if
       `frame-ancestors` contains a wildcard host.

remediation:
  manual: |
    1. For each Dataverse environment, browse to **Manage** > **Environments** > select it >
       **Settings** > **Product** > **Privacy + Security**. Role: Power Platform Administrator, or
       System Administrator of the environment.
    2. Turn on **Enable reporting** with a reporting endpoint, and leave enforcement off for a few
       days to collect violations (report-only mode).
    3. Add the legitimate embedding hosts to `frame-ancestors` (for example
       `https://teams.microsoft.com/` for Teams integration).
    4. Turn on **Enforce content security policy** on the **Model-driven** pivot, then on the
       **Canvas** pivot, starting with test environments.
    5. Equivalent: set the organization columns `iscontentsecuritypolicyenabled` and
       `iscontentsecuritypolicyenabledforcanvas` to `true` through the Dataverse Web API.
  rollback: |
    Turn **Enforce content security policy** off for the affected app type; keep reporting on.

references:
  - title: Content security policy
    url: https://learn.microsoft.com/power-platform/admin/content-security-policy
    publisher: Microsoft
    kind: vendor-doc
    accessed: 2026-10-24
  - title: CISA SCuBA Power Platform baseline
    url: https://github.com/cisagov/ScubaGear/blob/main/PowerShell/ScubaGear/baselines/powerplatform.md
    publisher: CISA
    kind: standard
    accessed: 2026-10-24
mappings:
  - framework: cisa-scuba-m365
    framework_version: "2.0.0"
    id: MS.POWERPLATFORM.4.1v1
    relation: equivalent
  - framework: nist-800-53r5
    framework_version: "5.1.1"
    id: SI-10
    relation: related
  - framework: nist-800-53r5
    framework_version: "5.1.1"
    id: SC-18
    relation: related
threats:
  - framework: mitre-attack
    id: T1189

authors:
  - name: Alice
    github: Sn0wAlice
introduced_in: 0.1.0
created: 2026-10-24
updated: 2026-10-24