Evidence intelligence
Verification infrastructure
Sign in ↗
R/IV / DOCUMENTATION

BUILD WITH
BOUNDARIES.

An integrated documentation surface with quick-start principles, primitive semantics, evaluation outcomes, and production boundaries.
AudienceAgent builders
Public accessConcepts and documentation
ExecutionAuthenticated and gated
00 / DOCS NAVIGATOR

Concepts should be inspectable.

Explore the design of the core verification contract with clear boundaries and direct documentation destinations.

DOCUMENTATION / FOCUS

Declare the consequential action.

A verification request identifies the proposed claim or action, supporting sources, authorization constraints, and scope.

Open reference ↗
INTERACTIVE SYNTHETIC DEMONSTRATION — NOT A LIVE CUSTOMER CASE OR PRODUCTION EXECUTION R/IV / BESPOKE EXPERIENCE
01 / QUICK START

Start with consequence.

Decide which actions demand prior checking: publishing a claim, attaching a citation, committing a transaction, or invoking an external tool. Give the verification layer the declared evidence and limits it needs to produce a bounded answer.

  1. Define intended action and its consequence.
  2. Specify source locators, expected claims, authority and allowed spend.
  3. Check support, conflict and uncertainty states.
  4. Use the result to stop, request review or execute.
  5. Record a receipt that can be inspected later.
02 / BOUNDARY

Public is not private.

The public website explains the method and bounded demonstrations. Customer records live behind authentication. Engine Core providers, owner controls and private evidence ledgers must remain inaccessible from the public interface.

03 / PATTERN

Verify before act.

The pseudocode below is an illustrative pattern, not a runnable client library. Production integration must use the exact endpoints, schemas, identity and payment modes exposed by the live deployment.

R/IV / EXAMPLE / CONCEPTUAL
// Illustrative integration pattern — not a live executable SDK example
const intent = agent.proposedAction();
const result = await riv.verify({
  action: intent,
  sources: context.sources,
  policy: authority.scope,
  spendLimit: budget.limit
});
if (result.status !== "verified") return stop(result.limitations);
await agent.execute(intent);
await riv.receipt(result);
04 / STATUS

Availability is explicit.

Do not infer executable entitlement from discovery metadata. A production machine action should pass authenticated identity, scope limits, applicable payment or entitlement, available provider execution, result validation, receipt issuance and replay safety before it is described as autonomously fulfilled.

Consult the current public R/IV production surface for live claims. This preview makes no operational promise about any individual execution endpoint.