---
schema_version: "newruntime-agent-readable-v0.1"
type: "trend_pattern"
slug: "harness-architecture-outlives-model-choice"
title: "Harness architecture outlives model choice"
description: "Rules, tools, permissions, memory, evaluation, execution, and recovery are becoming a stable runtime layer while models remain replaceable components."
thesis: "For production agents, the harness is becoming a more durable product boundary than the identity of the model running inside it."
status: "published"
confidence: "high"
first_seen: "2026-06-28"
last_verified: "2026-07-16"
supporting_signals: ["tg-2708","tg-2692","tg-2678","tg-2582","tg-2567","tg-2607"]
related_posts: ["agent-api-design","claude-code-subagents-memory","hilos-development-control-plane"]
counter_evidence: ["Provider-native harnesses can exploit proprietary model behavior and product features that portable runtimes cannot reproduce.","Model changes still alter tool use, planning, latency, and safety enough that switching is not operationally free."]
revision_trigger: "Revise the thesis if model-provider coupling consistently produces better reliability and lower total migration cost than portable rules, tools, traces, and execution contracts."
topics: ["harness-engineering","agent-runtime","model-routing"]
source_urls: ["https://opencode.ai/docs/models","https://github.com/NousResearch/hermes-agent","https://github.com/langchain-ai/deepagents","https://github.com/rivet-dev/agentos","https://docs.inference.net/introduction"]
routes: {"html":"https://newruntime.com/patterns/harness-architecture-outlives-model-choice/","markdown":"https://newruntime.com/patterns/harness-architecture-outlives-model-choice.md","json":"https://newruntime.com/patterns/harness-architecture-outlives-model-choice.json"}
source_format: "markdown"
---

# Harness architecture outlives model choice

## What is stabilizing

Agent products repeatedly need the same non-model capabilities: project rules,
tools, permissions, execution isolation, memory, context compaction, subagents,
traces, evaluation, and recovery. OpenCode makes providers swappable inside one
developer experience. Hermes combines a model endpoint with skills and tools.
Deep Agents and agentOS package orchestration and execution as reusable runtime
components.

The model still matters, but it is no longer the whole system.

## Supporting evidence

- OpenCode preserves rules, skills, sessions, and commands across model
  providers.
- An inference gateway gives applications a stable access contract while
  routing policy changes behind it.
- Deep Agents packages planning, files, memory, middleware, and subagents as a
  reusable harness.
- agentOS treats isolation, durable workflows, limits, and host-side secrets as
  runtime responsibilities.

## Operational consequence

Teams should version the harness contract separately from model configuration.
That means portable task fixtures, explicit tool schemas, provider-independent
traces, and a regression suite that runs whenever either the model or runtime
changes.