Arcane.localAI

This focused page is derived from the mechanically checked full member inventory.

Syntax

Arcane.localAI

Member kind

Namespace

Description

Native-host local-AI readiness, reconciliation, recovery, and isolated inference.

Overview

Arcane.localAI is the policy-filtered local-AI control surface. It covers readiness, Provisioner reconciliation, admitted service recovery, verified parallel-request configuration, and an application-owned isolated-model inspection and question lifecycle. Application selectors should use its admitted catalog rather than raw Ollama inventory.

Availability and security

Status requires ai.inference; lifecycle methods add app-id, app-type, privilege, host, and exclusive-mutation restrictions. Parallel-request mutation is Core-only, privileged, requires ai.runtime.manage, and is admitted only to the owning application. Isolated operations are restricted to their owning application. Long operations can emit operation.*; isolated questions correlate localai.isolated.phase by operationId.

Example

const access = await globalThis.Arcane.capabilities.list();
const canInspectLocalAI = access.methods.includes('localai.status');

console.log('Policy-filtered local AI status admitted:', canInspectLocalAI);

Members

• Arcane.localAI.status() — Returns the native host's authoritative package-policy-filtered catalog and readiness. Desktop Core includes native resource and speech admission. An admitted Android application host may instead return providerMode:"user-managed-loopback"; that status has no managed lifecycle, but it merges bounded native Kokoro/Whisper role health and catalog evidence while Android speech operations use the separate Arcane.speech bridge. Requires ai.inference.
• Arcane.localAI.ensurePlatform() — Runs the Provisioner-only post-install AI reconciliation as one exclusive operation: returns an explicitly selected, configured OpenAI state without touching Ollama, or hardware-admits, ensures, and loads the managed local model according to boot settings. Requires provisioning.manage and the provisioner app type.
• Arcane.localAI.recover() — Requests one privileged, exclusive, start-existing-service-only recovery of selected Arcane-managed local services. Requires ai.inference and explicit application admission.
• Arcane.localAI.setParallelRequests() — Solves one app-owned verified model/context's maximum directly from native model and memory/resource evidence without a product ceiling, clamps a positive request to that result (0 requests the maximum), and changes the machine-wide managed Ollama setting only when needed. A tentative change uses one fresh stopped-service snapshot before the final value is written, then Core ensures, loads, and confirms the exact model. A restart unloads resident models. Automatic mutation currently requires the dedicated ai.runtime.manage capability on Microsoft NT desktop Core and operating-system authorization; Linux returns administrator-managed systemd guidance, and Android user-managed loopback sessions do not invoke it.

Reference group

Namespace, constructor, and values