Security
Keeps intruders and scams out.
Whatever could let someone in – a sign-in from a new place, a hole in a device or in code, a leaked key or spoofed mail, for example – is found and followed to its cause, and Xovris closes it itself, saying what it changed.
XovrisOn its own
- A lit square is a feature people use today
- Newest: The AI review runs on every paid plan
Every feature in Security, strongest first.
What fixes and prevents a problem comes first. Type to find one, or open a plan to see only its own.
Nothing here matches those words yet. Xovris notes the words, with no address and no cookie, so what is missing gets written.Ask Xovris instead
Lit: live and workingOpen: Xovris is still working on it
The AI review runs on every paid plan
Every paid plan has the adaptive AI review of changes; on the free plan Xovris's own checks keep watch, and the plan says so.
Your code's bill of materials in the command centre
A connected repository's review shows its packages with their licences and advisories, and which of them are in a build that is live.
See what people did beside what Xovris did
Sign-ins, approvals, member, token, webhook, autonomy and settings changes join Xovris's own actions in one searchable record, under 'What people did'.
Know which vulnerable packages are actually served
Xovris lists every package in your code with its licence and advisories, and says which vulnerable ones your live site actually serves.
See who may do what at a glance
The access page opens on a grid of every real grant, person by system, with a risky grant ringed in red beside what Xovris saw and the fix.
Password health across the company
Xovris finds reused or leaked passwords across your company and moves people to passkeys.
Travel mode for laptops and phones
Before someone travels abroad Xovris locks their laptop and phone down, and checks them again on return.
Insider risk, lawfully and openly
Xovris spots data being copied out or access being misused from inside, under lawful and transparent rules.
Your data masked before any AI reads it
Secrets and personal details are masked before any AI reads your records.
Secrets and personal data masked, field by field
Before anything is stored or analysed, secrets and personal data are masked, and the masking is proved for each field.
Alerts when a key is used in a way it never was
If an API key is used from a new network, at a new hour or in a sudden burst, Xovris alerts you.
What you delete stays deleted
A fact you delete is gone from everything Xovris reads, and stays gone even after a restore from an older backup.
Containment lifts itself when it is safe
Every containment Xovris puts in place carries its own exit test: it lifts when the threat is gone and holds, with its reason, while it is not.
Containment checked where the threat was
After containing something, Xovris checks at the source that the suspicious activity has stopped, and says plainly if it has not.
A containment is judged by what it stopped
A containment's verdict comes from watching for the threat again, never from the tool saying it acted.
Quarantine lifts only once it is clean
A quarantined device, a disabled account or a blocked address is released only once the threat and anything it left behind are gone and the system reads healthy.
Containment rehearsed before it is used
Every containment is tried on a copy first; one that would not work or would take a service down is refused before the real system is touched.
Containment never takes your other services down
Before and while a containment is in place, Xovris checks every other service the system serves, and undoes it at once if one stops answering.
Sign-in problems explained from the source
When someone cannot sign in, Xovris reads your identity provider's own sign-in diagnostics and says what went wrong and what to do.
Session replays masked by default
Masking is on by default in every session replay, proved on emails, card numbers, phone numbers and national insurance numbers.
Feedback reaches its own session and no other
A visitor's note travels with their session's signals and is linked to that session alone.
Masking proven everywhere a session is seen
Masking holds in replays, diagnostic screenshots, logs and anything handed to an AI.
Identity steps done for you
When an identity check hands you a step, Xovris carries it out itself through your connected directory.
Security headers set on any web server
A site on a server Xovris has no built-in connection for gets its six security headers written in that server's own configuration.
An approval runs only the exact change you approved
When you approve a repair, only the exact package of changes you saw runs - anything different needs your yes again.
Vulnerabilities read from the authority's own records
Xovris checks your software against real OSV advisories across npm, PyPI, Maven, Go, crates.io, NuGet and more, measured against the authority's own answer.
One emergency stop for everything
One stop halts every automatic action Xovris could take, checked against what the running code really does.
Every access decision explained in plain words
Each time Xovris allows or refuses something, it tells you in plain words what was asked and why.
Payment requests by voice or video held for checking
A request to pay that arrives by a voice, a face or a chat claiming to be someone is held until it is confirmed another way - how deepfake fraud is stopped.
Every action from chat is re-checked at the moment it runs
An action asked for in Slack, Teams, SMS, WhatsApp, email or by phone is authorised against the person again at the moment it runs.
Xovris acts only on messages from people still in your team
A request from someone who has been removed - by Slack, Teams, SMS, WhatsApp, email or phone - is refused before anything runs.
Mail records checked beside every DNS repair
Every DNS repair reads your MX, SPF, DKIM and DMARC beside it, so fixing one record never breaks your mail.
DMARC read to the newest standard
Xovris reads DMARC to RFC 9989 and 9990, including the newer settings for subdomains that do not exist.
Every inbound callback authenticated
Every callback your own tools send to Xovris is authenticated, and a captured delivery sent again is answered but never acted on.
Inbound events checked against replays
Every event your tools send to Xovris is checked as it arrives, and a replayed, re-ordered or out-of-date delivery is turned away.
Chat replies act as the person who sent them
A reply from chat, text or WhatsApp acts as that person's own Xovris member, with their role read at the moment it acts.
Xovris acts only for people in your team who may act
Xovris runs a chat command only for someone it knows in your team who may act. A sender it does not know, who has left, or who only views gets an answer and nothing runs.
A DMARC policy of 'none' moved on for you
Once SPF, DKIM and DMARC are read, a DMARC policy left at none is tightened by Xovris through your DNS connection.
Nothing you send is ever used to train AI
Xovris works only with AI whose own terms say it never trains on what is sent, so nothing you send Xovris teaches anyone's AI.
New tools run sealed off the first time
A tool Xovris fetches runs its first time in a micro-VM of its own, with no network and nothing written outside it.
Your data walled off from every other workspace
Every table that holds a workspace's data is confined to that workspace by row-level security the database itself enforces.
Every action Xovris takes is on a tamper-evident record
Every act Xovris takes on your systems joins your workspace's append-only, hash-chained record in the same transaction as the act itself.
Everything Xovris is built from is pinned
Every GitHub Action Xovris uses is pinned to its commit and every image to its digest, so nothing it is built from can change underneath it.
Decoy accounts and documents that catch intruders
Xovris places decoy accounts, shares and documents in your systems and tells you the moment anyone touches one.
Expiring keys rotated before they lapse
Where a provider's own API issues a new key, Xovris rotates an expiring key itself, retires the old one and tells you.
The check and the fix say exactly the same thing
The email check on xovris.ai runs the product's own mail check, so the problem it names and the fix Xovris writes are always one and the same.
Any security product connects, not only the five listed
A security product Xovris has no built-in connection for is connected from its own API description: its devices, detections and weaknesses read, its actions built, and every undo proved on a copy first.
Every repair checked against the access you granted
Xovris reads back what each connected service actually lets it do, from the service itself, and checks every repair against it before offering it.
Every change passes one safety gate
Every kind of change Xovris can make - including in a browser it drives and in a service of yours - passes the same safety gate first.
Nothing is lost, and backups are proven by restoring them
A change is made only once what it touches is backed up, and the backup is proven by restoring it into a copy. A secret field's value is never read.
Certificate fixes rehearsed with a real certificate authority
A certificate fix is rehearsed in Xovris's sandbox with a real ACME order, its chain verified, before your site's certificate is touched.
Apps with unusual access to your data, found
Every app people have let into your Google Workspace or Microsoft directory is read with the access it holds, and the unusual ones are flagged.
Single points of failure in people, found
Xovris reads your directory and finds who alone can administer, who signs in with a password alone, shared logins and accounts of people who have left.
Weak or missing DKIM keys found at your mail provider
Xovris reads your DKIM keys from your connected mail provider and finds any that are missing, revoked or shorter than 2048 bits.
DNS records left pointing at nothing are removed
Xovris finds DNS records still pointing at something you no longer own - the way subdomains get taken over - and removes them, with a way to put each one back.
A certificate that does not match is reissued
A certificate that does not cover the name your site serves is reissued with the right names, installed and checked from outside, and an incomplete chain is put right.
Lost your authenticator? Xovris hands you the way back
If you say your phone or authenticator app is gone and a system then asks for a code, Xovris gives you that provider's own recovery page on the same job.
A locked account is never made worse
If a system says your account is locked, Xovris gives you the provider's official unlock and sends that system nothing more until it is unlocked.
Tool searches carry nothing of yours
When Xovris searches for a tool, the search names only the kind of job. Nothing of your workspace ever leaves with it.
Every tool's licence is read first
A tool with no licence, a non-commercial licence or one that forbids the use is refused before it runs.
Abandoned and look-alike packages are refused
A withdrawn, long-silent, days-old or look-alike package is refused, and Xovris shows the reason beside the evidence it read.
Tools with known vulnerabilities are refused
The exact version of every tool is checked against the vulnerability database before use. One advisory refuses it, and a database that does not answer is never read as clean.
Every tool is checked byte for byte
The bytes Xovris receives must match the tool's published checksum. A single changed byte refuses the tool.
Tools are removed when the job is done
A tool fetched for a job is checked again as it lands, used for that job only and removed afterwards, with nothing left behind.
The other areas
Words on this page, in plain English
Each technical word this page uses, said simply.
- A message sent to your own system (webhook, webhooks)
- An automatic message one service sends to an address you choose the moment something happens.
- Program-to-program connection (API, APIs)
- A way for one program to ask another for information or to do something, without a person clicking through a website.
- Domain Name System (DNS)
- The internet’s address book: it turns a name you type into the number of the computer that answers for it.
- Allowed senders record (SPF)
- An address-book record listing the servers allowed to send email as the domain.
- Email signature (DKIM)
- A signature added to each email that proves it came from the domain and was not changed on the way.
- Email rules record (DMARC)
- An address-book record telling mailboxes what to do with email that fails the domain’s sender checks – deliver it, mark it as spam or refuse it – and where to send reports.