ONERINGAI V2

Documentation

Guides, API reference, and examples for oneringai v2.

RUNTIME WALKTHROUGH

AMOS cookbook ↗

Set up a host, run each driver, inspect activity, and resume sessions.

Beta 2.0.0-beta.1 · Docs bebdcd1 · Updated 2026-11-02. External harnesses are preview; provider access is required.

Get started

Quickstart ↗

Install the beta and run a first agent.

User guide ↗

Agent setup, tools, context, and persistence.

API reference ↗

Public types and method signatures.

Examples ↗

Runnable recipes with offline and live modes.

V2 controls

V2 adoption guide ↗

Features, defaults, support, and limitations.

Durable evidence ↗

Retained results, projections, and recovery.

Task state ↗

Host-confirmed constraints and action receipts.

Helpers & budgets ↗

Helper models and shared execution limits.

Integrations

Connector catalog ↗

Authentication templates and tool packs.

Memory system ↗

Entities, facts, relationships, and retrieval.

Agent runtime ↗

Definitions, sessions, harnesses, and events.

Upgrade from v1 ↗

Move to the unscoped oneringai package.

All guides

Agentic baseline evaluation — Stage 00 Bounded recovery and host completion checks Agent Orchestrator Architecture Plan Opt-in Agent presets and capability inspection Whole connector collection pages Connector & Tool Catalog Consumer capacity inventory Selective context assembly and stable prompt layout Context preparation and freshness Opt-in context strategies (Stage 07)Published dependency review Durable tool evidence (Stage 03)Endpoint capability companions Enhanced connector contracts Evidence conformance and composed guarantees Logging Configuration Memory Layer — API Reference Memory Layer — User Guide Memory Permissions — Usage Guide Memory Predicate Library — Usage Guide Memory Signal Ingestion — Usage Guide Historical v1 upgrade notes Model Registry and Provider API Audit — 2026-10-07 Managed model routing Model tasks and purpose-specific helpers Plugin and provider-native context compaction Complete host configuration for orchestrated agents Invocation-owned tool failures v1 to the unscoped OneRingAI v2 package Effective parameter assessment and temperature policy OneRingAI v2 release readiness Negotiated remote tool outcomes Protected task state Tool outcomes and cancellation Vendor presentation without identity changes Verified routines (C3)OneRingAI Version 2 Guide

Documentation snapshot · 2026-11-02Suggest a correction ↗