Stable Drift Detection

What you declared should equal what is deployed.

Authority drifts quietly. A config change here, a redeploy there, and a deployed AI agent now holds power you never approved. Stable Drift Detection watches for exactly that. It continuously compares deployed authority against your declared state. The moment they diverge, it alerts you. You find the gap in minutes instead of during an incident weeks later.

Start a proof of concept →See Reality Coupling™

Stable Drift Detection, point by point

Demo data

You learn: The page's points, one at a time. You can: Step through them.

  1. Step 1 of 5

    KYE Protocol™

    Agent authority

    The authority an agent holds in production versus the authority you declared.

  2. Step 2 of 5

    KYE Protocol™

    Policy bindings

    The rule packs actually loaded versus the rule packs the manifest names.

  3. Step 3 of 5

    KYE Protocol™

    Surfaces and endpoints

    The operations live in production versus the operations declared in the spec.

  4. Step 4 of 5

    KYE Protocol™

    Evidence chains

    Whether the §0.3 evidence family is still emitted on every privileged action.

  5. Step 5 of 5

    KYE Protocol™

    Disabled controls

    Gates or guards that were turned off without an approved change.

How this widget is built
Demonstrates
KYE Protocol™
Components
  • The points this page makes
Flow
Each point in order, as the page sets it out. Architecture diagram
Used on

The problem in one sentence

Your declared policy and your deployed reality slowly drift apart, and nobody notices until it costs you.

KYE Protocol™ identified this as the root of eight separate audit findings. A manifest declares an authority. A deploy quietly grants more. A gate exists to check it, until someone disables the gate. Each gap is silent. Stable Drift Detection closes the silence. It treats divergence between declared and deployed state as a first-class signal, not a surprise.

What it watches

Stable Drift Detection covers the full set of classes the Reconciliation Engine™ reconciles, not just one.

  • Agent authority. The authority an agent holds in production versus the authority you declared.
  • Policy bindings. The rule packs actually loaded versus the rule packs the manifest names.
  • Surfaces and endpoints. The operations live in production versus the operations declared in the spec.
  • Evidence chains. Whether the §0.3 evidence family is still emitted on every privileged action.
  • Disabled controls. Gates or guards that were turned off without an approved change.

How it works

Stable Drift Detection runs on the Reconciliation Engine™, KYE Protocol™'s declared-versus-deployed primitive.

The engine asserts one thing for every artefact: declared state equals deployed state. It runs on events, not on a clock, so a change is checked when it happens. This add-on subscribes to its findings. It scores each divergence for severity. It routes the urgent ones to you and logs the rest. Every run seals its own evidence, so the check is itself auditable.

The value for your team

Stable Drift Detection turns a class of silent failures into early, actionable alerts.

  • Catch it early. Find a drifted authority in minutes, not during a post-incident review.
  • Fewer incidents. A divergence flagged at deploy time never becomes a production breach.
  • Audit confidence. You can show a regulator that declared and deployed state are kept in lockstep.
  • Less manual checking. The engine watches every class continuously, so your team stops spot-checking.

How it is sold

Stable Drift Detection is an annual add-on to a KYE Protocol™ platform tier.

The fee is a hybrid runtime plus annual renewal charge. Stable Drift Detection is also a member of the KYE™ Advanced Assurance Bundle™, which packages the four advanced add-ons at one price. Pricing is shared by email under our closed-registration policy.

Start a proof of concept →See the Advanced Assurance Bundle™

Canonical KYE™ surfaces referenced on this page: KYE Protocol™ · KYE™ Reconciliation Engine.