Security overview
Autonomy, with every control in your hands.
How Rexship acts on its own, what it can touch, and how you govern and stop it. For a full review, write to us and we will answer in writing.
How autonomy is governed
Rexship acts on its own only inside the policies you set. A policy decides which kinds of change ship on their own and which wait for a named approver, and it names who that approver is.
Budgets and change windows set how much Rexship may spend and when it may release. Sensitive actions need two-person approval, and every approval is attributed to the person who gave it.
What Rexship can touch
Only the repositories, services and environments you connect. Each connection uses least-privilege credentials scoped to that system, and you can revoke any of them at any time.
Every organization runs inside its own isolated boundary. Your code, your data and what Rexship learns about your systems are never shared across customers.
How work is contained
Builds and tests run in sealed, throwaway environments with no network and no standing credentials, and the environment is discarded when the run ends.
Every model output passes a safety check before it can become a change or an action. Changes then roll out in stages through your gates, with automatic rollback when health checks fail. Traffic reaches your services through Rexship’s edge, with filtering and rate limits in front of every endpoint. Dependencies are scanned continuously, and known-exploited vulnerabilities are patched first. Every push and pull request is reviewed for eight kinds of weakness: security, correctness, reliability, drift from intent, tests, hallucinated APIs, inert code, and style and consistency.
Your code and data
Your code and data are used only to build, run and maintain your software. They are never used to train models.
Code, secrets and logs are encrypted in transit and at rest. Run on Rexship’s cloud or in your own VPC with customer-held keys, in the regions you choose. Export and erasure are available on request.
The record
Every action Rexship takes, and every approval a person gives, is attributed and written to an append-only, hash-chained audit log that you can export.
Each export carries the head of the chain, so a log that has been trimmed or altered is detectable.
- Change
- Patch a vulnerable dependency in the checkout service
- Raised by
- Rexship · nightly scan
- Build
- Sealed environment · passing
- Your suite
- Before the change: passing · with the change: passing
- Regression test
- Added · fails before the fix, passes after
- Policy
- Routine security patch · ships on its own
- Release
- Staged rollout · health checks passing
- Record
- Attributed · written to the audit log
Stopping Rexship
Pause Rexship for one service or for everything, at any time. The stop is yours to pull, and so is the decision to resume.
Identity and access
Single sign-on for your people, and roles at organization and service scope, so each person sees and approves only what their role allows.
Reporting a vulnerability
If you believe you have found a vulnerability in Rexship or in this site, write to admin@rextail.ai with “Rexship — security report” as the subject.
We acknowledge your report, keep you informed while we investigate, and coordinate disclosure with you. Please give us a fair chance to fix the issue before you share it publicly.
Questions for your security team
We answer security questionnaires in writing, and we will walk your security team through the architecture and the controls described here.
Security overview · Revision A