- Sign-in
- Passkeys
- for every person, with SAML and OIDC single sign-on on Premium
- Agent changes
- Review
- require a person’s approval before screens change, on Premium
- Audit log
- 400 days
- changes to screens, playlists, schedules, people and agents, chained by hash
- Players
- Native
- cards drawn by the Player, and every file checked by SHA-256 before it plays
People sign in as themselves
Nobody shares a login. Each person has their own sign-in, a role in each project and, where you choose, a set of screens they are limited to.
Single sign-on (Premium)
People sign in through your identity provider with SAML 2.0 or OpenID Connect, and Microsoft Entra ID has a preset. Require single sign-on and passwords and passkeys stop working for your verified domains, apart from up to two break-glass admins. Single sign-on guide
SCIM provisioning (Enterprise)
Your identity provider adds and removes people. Deactivating a person your organization manages ends their sessions, suspends their project access and suspends the agents they are responsible for.
Passkeys and passwords
People sign in with a passkey or a password. A password has at least 12 characters, is refused when it is a common one, and is stored as an Argon2id hash. Sign-in attempts are limited for each email address and each IP address.
A fresh sign-in for sensitive changes
Changing a password, a passkey or an email address, approving or disconnecting an agent, and changing admins, the identity provider or SCIM each ask for a password, a passkey or a recent single sign-on again.
Roles
A project has an Owner, Approvers, Members and Uploaders. An organization has admins, members and external guests. An Uploader (Plus) adds only to the playlists and media folders you choose, and never approves.
Screen folder scopes (Premium)
Limit a person or an agent to chosen screen folders and screens. Everything outside is hidden and unreachable, and a changed scope applies from the next request. Permissions guide
Every change has an owner and a record
Decide who may change a screen, who must approve it, and read afterwards exactly what changed.
Review of agent changes (Premium)
Turn on Require review for agent changes and an agent’s playlists, screen assignments, schedules and takeovers wait for a person who can approve. An agent never approves a change.
Protected folders (Enterprise)
A change that reaches a protected folder’s screens needs a second approver: someone other than the person who proposed it, and other than the person responsible for the agent that proposed it.
Audit log
Changes to screens, playlists, schedules, people, agents and settings are recorded with the person or agent that made each one, who approved it, the values before and after, and where the request came from. Entries are chained by hash, so an altered record shows, and are kept for 400 days on every plan. Reading and exporting the log as JSON, CSV, XLSX or PDF is part of Premium.
Playlist revisions (Premium)
Every playlist keeps its revisions. Restoring an earlier one copies it forward as the newest revision, so nothing is lost.
Schedule change control (Enterprise)
Schedule entries start as drafts, play as a test, and reach every screen only when they are promoted.
Sign-in activity
Sign-ins, refused sign-ins and credential changes appear in the project’s activity log, and a person is emailed when their password or passkey changes.
An agent holds its own, limited access
An agent never borrows a person’s password. It has a credential of its own that says what it may do, and a person can take it away.
Capabilities fixed at issue
Each agent credential carries a fixed set of capabilities (screens, content, playlists, advertising, reports and project) and an access level: read, propose or manage.
Approved by a person
An agent that joins an existing project waits for a person to approve it in the dashboard with a fresh sign-in. That person can narrow what it may do, and never widen it.
Never more than its person
An agent limited to screen folders (Premium) reaches only what both it and the person responsible for it can reach.
Disconnect at any time
The dashboard lists every agent. Disconnect one and its credential is revoked. Deactivating a person suspends the agents they are responsible for.
Short-lived tokens
An agent connected with OAuth uses access tokens that last 10 minutes. Refresh tokens rotate, and reusing an old one revokes the whole grant.
Service clients
Your own services sign in with a key or a secret and receive tokens that last 15 minutes.
Screens that play only what you published
A screen has an identity of its own, receives only its own project’s content and checks each file before it plays.
Pairing
A screen pairs with a six-character code that expires, and five wrong tries lock a code for a minute. The code is never a credential: a native Player then proves itself with its own Ed25519 key.
Content for paired screens only
Your media, application packages and application data are delivered only to a paired screen’s session or a signed-in member. A screen reaches its own project’s media and the organization Library its project uses, and nothing else, so a file’s address alone opens nothing.
A screen leaves only at the screen
Nothing done from an account unpairs a screen. A paired screen leaves only when someone chooses Reset this screen on the device. From the dashboard you archive a screen, which stops its playback until you restore it.
Verified content
Every Player checks the SHA-256 hash and size of each media file, card and application package before it plays.
Native Players
On Android, the Amazon Signage Stick, Linux, Raspberry Pi, Windows, macOS, Apple TV, iPhone and iPad the Player is a native app. It draws cards itself, with no web view, and keeps the screen’s credentials in native code, away from web pages and applications.
Signed updates
Android and Apple Players update through their stores or your device manager, and the Windows Player through the Microsoft Store. The Linux and Raspberry Pi Player installs and updates only from a release manifest signed with an Ed25519 key, and checks each archive’s SHA-256.
Device keys
Android seals the screen’s key with a key held in the Android Keystore, Apple TV keeps it in the Keychain and Windows protects it with DPAPI. A Raspberry Pi image derives it from the card and the board, so a card moved to another board is a new, unpaired screen.
Menu PIN and kiosk lock
On Linux, Raspberry Pi and Android, set a PIN of 4 to 8 digits on the Player’s menu; after four wrong tries the wait grows from 30 seconds to 15 minutes. The Raspberry Pi image and the Linux kiosk install start the Player alone on the display, a managed Android device runs it in lock task mode, and Windows locks it with Assigned Access or Shell Launcher.
Applications, privacy and your data
What an application on a screen can reach, what a screen sends back, and what happens to your data.
Isolated applications
Each application release runs on its own origin under apps.screenrig.ai, with no account cookie. It talks to the Player through a message bridge, and through it reaches only what the Player grants. Uploaded packages are unpacked in a locked-down container with no network.
Camera and microphone
The SDK’s barcode scanning, people count and sound level are analysed on the device, and pass the application only the result: the scanned text, a count or a level.
Screenshots and live view
A screenshot or a live view is taken only when someone in the project asks for it, and PIN entry is never captured. Live view (Premium) is video without sound, and on Android a person at the screen allows it first.
See every screen
Screenshots, blank and frozen screen detection, and email and webhook alerts show what is on the glass and tell you when it is wrong, and live view (Premium) shows it as video.
Your data stays yours
screenrig project exportwrites a project’s configuration, original media and event log to your disk at any time. A deleted project’s data is removed after a 90-day recovery period, as the privacy policy describes.Secrets at rest
Agent, SCIM and alert trigger secrets are stored only as SHA-256 digests. Secrets screenRIG must use for you, such as identity provider, webhook and connected account secrets, are sealed with AES-256-GCM and bound to their project.
Encrypted connections
Every connection to screenRIG uses HTTPS, and browsers are told to keep using it with HSTS. The Android Player asks for no location permission and keeps its data out of device backups.
screenRIG staff
Staff sign in with passkeys only. A staff member who works in your project states a reason before any change, and each change is in your audit log.
Clean logs
The service and every Player remove credentials from logs and error reports before they are stored or sent.
The details
The smaller controls a security review asks about.
- Signed webhooks. Each delivery carries an HMAC-SHA256 signature. A webhook goes only to an HTTPS address on the public Internet, the address is checked again before every delivery, and redirects are never followed.
- Approved destinations. A webhook, a chat or pager notification and an event subscription deliver only after an owner or admin of the organization approves the exact address with their passkey. An agent can create one; only a person sends your notices to a new address. A delivery that carries a credential or personal data is not sent.
- Checked uploads. Media is PNG, JPEG, WebP, GIF, MP4, WebM or MP3. The declared type must match the file’s own bytes, and its SHA-256 and size must match what was declared.
- Rate limits. Sign-in, pairing, uploads, screenshots and screen control each have their own limit, and when one of those limits cannot be checked the request is refused.
- Sessions. A password session lasts 24 hours and a single sign-on session 24 hours. A passkey session lasts 7 days from its last use. Setting a password or adding a passkey ends a person’s other sessions.
- Dashboard hardening. The session cookie is HttpOnly, Secure and SameSite=Strict, and a content security policy allows only the dashboard’s own scripts.
- Fleet links. A fleet link enrolls screens with no pairing code. Give it an expiry date and a maximum number of screens, and disable or rotate it at any time.
- Expiring links. A sign-in reset link lasts 1 hour and an emailed invitation 7 days.
- Emergency alerts (Premium). A CAP alert trigger has its own secret, stored as a digest, and cannot target a protected folder’s screens.
- Safe exports. CSV exports prefix any cell that starts with a formula character, so a spreadsheet never runs it.
- Report a vulnerability. Tell us through the contact form. The CLI and plugin repositories each carry a SECURITY.md with their private reporting path.
Questions from your security team?
Send your questionnaire through the contact form. Each answer says how the control works.