masood-2026-agentic-harness-engineering · post

Agentic Harness Engineering: The Operating System for Machine Intelligence

Adnan Masood — Medium

Created: 2026-06-27 · Ingested: 2026-09-11

https://medium.com/@adnanmasood/agentic-harness-engineering-the-operating-system-for-machine-intelligence-ce4fa48e9520

A 37-minute practitioner's map of agent harness design — boundary, loop, context, tools, memory, verification, orchestration, reliability, safety, economics, measurement, trajectory — drawn, by its own account, from the public engineering record of Anthropic, OpenAI, Cursor, Cognition and OpenHands. Its headline figure: hold the model fixed and swap only the scaffolding, and an identical Claude Sonnet 4.5 moves from 30.91% to 74.55% on GAIA.

Referenced by

Claims in this catalog that draw on this source, and whether as support or counterpoint.

supportsA large share of an agent benchmark score is attributable to the harness around the model rather than the model itself: holding the model fixed and changing only guides, sensors, retries and state management is reported to move the same model from 30.9% to 74.6% on GAIA and 30th to 5th on Terminal Bench — so a score reported for "a model" on an agentic benchmark is a score for a model-plus-harness, and comparisons across labs with different harnesses are not comparisons of models.