Unit Tests Passed
Across 111 test files. Verified main build, 4 November 2026.
Latest verified main build: 040ce16 · 4 November 2026. Build, architecture, verification and No Means No proof gates passed. View build evidence ↗
Across 111 test files. Verified main build, 4 November 2026.
Across 4 test files against the Firestore emulator on a hosted runner.
5,000 assets assessed across an 84-day baseline and 420,000 evidence rows. 4,323 assets verified.
The controlled 5,000-asset proof recorded 1,157 authorised effective executions and durable receipts, with zero duplicates. Exactly-once applies to the tested contract and receiver observations.
Modelled electricity-cost exposure of A$616,678.14, calculated from 2,055,593.799 kWh at A$0.30/kWh.
Modelled emissions exposure of 1,336.136 t CO2-e, calculated at 0.65 kg CO2-e/kWh.
An instruction becomes an action only when permission is valid. KƎƎLO™ records the result and keeps unresolved work visible.
Intelligence and compute propose an operating objective. Models and agents do not acquire authority by proposing it.
Director coordinates operating priorities. Forge supports construction of bounded capabilities.
KƎƎLO™ checks the rules, current permission and operating state before acting.
OrganisationPolicyEvaluatorDelegatedExecutionAuthorityServiceGovernedExecutionGatewayApproved actions reach connected systems through controlled interfaces.
An enduring record shows what happened and what still needs attention.
Executed, denied and unresolved outcomes remain connected to their governing evidence.
The requested action, actual result and supporting evidence are checked against each other.
A person or model proposes an action. The request specifies the result it is trying to achieve.
The request enters a controlled path with the context needed to identify the organisation, actor and requested operation.
An authorised operation crosses a controlled adapter into the target system. Duplicate protection applies within the tested execution and receiver contract.
The receipt records the outcome available to KƎƎLO™ and ties it to the authority and evidence behind the request.
The system checks the observed result against the intended effect. An uncertain result remains visible for review.
The system compares the request, receipt and observed result. Any unresolved work stays open until it can be accounted for.