Security engineering, risk, and compliance
Answer the question before it is asked.
Autonomous systems are moving into scope for the frameworks you already report against. Cirvix produces the artefact those controls need — a tamper-evident record of what every agent did, under which policy, on whose authority.
Ask yourself
If an agent took a destructive action last Tuesday, how long would it take to produce the authorization record?
Today
What you are working around
- You cannot enumerate the agents in the estate, let alone their access
- Evidence for an agent-related control means joining three systems by hand
- Approval for sensitive actions happens in chat and leaves no durable record
- A model provider changes behaviour and you have no way to see what changed downstream
What changes
The specific difference
The record is the control
One append-only entry per decision carries agent, owner, action, resource, policy, verdict, and approver — so a control test reads one system.
AuditTamper-evident by construction
Each entry commits to its predecessor and checkpoints are published, so alteration after the fact is detectable rather than merely discouraged.
AuditApprovals become evidence
Human sign-off happens inside the execution path, so 'a person authorized this' is a record rather than a recollection.
Human ApprovalSeparation of duties that holds
The agent proposes, the policy decides, a named human approves. Those are three principals, and the log distinguishes them.
Policy Engine
In detail
What Cirvix does and does not give you
Cirvix produces evidence. It does not certify you against any framework, and no product can. What changes is that the evidence exists continuously, in one place, in a form that maps to controls — instead of being assembled under time pressure during fieldwork.
Tamper-evidence is not tamper-proofing
A hash chain proves records were not altered after they were written. It does not prevent someone with sufficient access from destroying the log, and it cannot prove a record was written truthfully in the first place — that property comes from the enforcement path, not from the log. We would rather you hear that from us than from your auditor.
Scope grows before the frameworks do
Guidance on autonomous systems is still settling, but the underlying questions are not new: who authorized this, what could it reach, and can you show me. Instrumenting the decision path now means the answer is already accumulating when the specific control language arrives.
Get started
Let the agents do real work.
Start with the free local scanner — it needs no account and tells you what is already ungoverned on your machine. Move to a governed runtime when you are ready.
Cirvix AgentControl is in private beta. Pricing is published and plans are not yet chargeable — early access confirms your place and we will tell you before anything bills.