What is mattwood.fyi?

A live, structured feed of riffs and links from Matt Wood, Chief AI & Technology Officer at AWS — readable by people and queryable by agents as a typed knowledge graph.

Is there an API and is it free?

Yes. A public, read-only query API is available with no authentication and no API keys. See https://mattwood.fyi/openapi.json and https://mattwood.fyi/docs.

How should an AI agent use this site?

Fetch https://mattwood.fyi/feed.json for the full structured feed, or query the API for keyword/semantic search and the connection graph. See https://mattwood.fyi/agents/ for when-to-use guidance.

Harness Engineering Paper Collection

A curated collection of 21 papers tracing the evolution of AI model harnesses—the infrastructure surrounding model weights—from basic sampling loops in 2019 to self-modifying systems in 2026, showing how the same model can achieve vastly different performance depending on its surrounding harness architecture.

academy.dair.ai →

Connections

Supports

Agent Memory as a File Format Agent Memory as a File Format addresses a core harness infrastructure concern—how persistent state is managed around model weights—matching the collection's focus on surrounding infrastructure evolution
Declarative Attention: Models Controlling Their Own Attention Declarative Attention describes models controlling their own attention mechanisms, representing the frontier of harness evolution toward self-modifying systems that the collection's 2026 papers would document
Domain-Driven Agents Domain-Driven Agents addresses the infrastructure and scaffolding surrounding model weights, directly within the harness engineering domain the collection surveys

Related to

Ornith-1.5: Self-Scaffolding to Self-Improvement Ornith-1.5 describes a self-scaffolding to self-improvement system, directly exemplifying the 'self-modifying systems' endpoint of the harness evolution timeline traced in the paper collection
vLLM v0.28.0 Release vLLM is a prominent inference and serving harness framework; its release notes reflect the ongoing engineering of sampling and execution infrastructure that the paper collection traces from 2019 onward
Munder Difflin: Digital Clone Agent Harness Munder Difflin is explicitly a 'Digital Clone Agent Harness'—a concrete instantiation of the harness infrastructure concept the paper collection documents and traces historically

Jalapeño Shows Power of LLMs for Chip Design Both relate to AI-assisted hardware/systems engineering workflows, with the Harness Engineering Paper Collection likely covering adjacent topics in AI for hardware design

Type:	link
First published:	Sunday, 13 September 2026 at 21:37 UTC
Location:	https://academy.dair.ai/papers/collections/harness-engineering
URL:	https://mattwood.fyi/i/4098d010

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Connections