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.

EnvHarness: Awakening Static Worlds for Agent Learning

EnvHarness is a programmable framework that dynamically adapts static environments for LLM agent training by wrapping and reshaping behavior without modifying underlying logic. EnvRigger automates this process by observing agent trajectories to synthesize targeted environment modifications that improve learning efficiency and performance across multiple domains.

arxiv.org →

Connections

Supports

Munder Difflin: Digital Clone Agent Harness Munder Difflin's digital clone agent harness is a concrete instantiation of the kind of agent harness pattern that EnvHarness provides infrastructure for, supporting the broader harness-for-agents paradigm
Bridging Intent and Execution in Agentic Systems EnvHarness addresses the gap between agent intent and execution by reshaping environment behavior, directly supporting the problem of bridging intent and execution in agentic systems
SWE-bench Science: Coding Agents for Engineering Tasks SWE-bench Science evaluates coding agents on engineering tasks, and EnvHarness's dynamic environment adaptation directly addresses the challenge of creating reproducible, controllable environments needed for such agent benchmarking

Related to

Every Model Cheats: Prompt-Level Mitigation of Cheating on Offensive Cyber Tasks Both deal with environment-level manipulation affecting agent behavior — EnvHarness wraps environments to shape training, while the cheating mitigation work shows how prompt/environment framing changes agent outcomes on tasks
DeepSeek Harness: Plugin Architecture DeepSeek Harness uses a plugin architecture for wrapping model behavior, directly analogous to EnvHarness wrapping and reshaping static environments without modifying underlying logic
Building an Advanced Agentic Harness Both focus on building agentic harness frameworks for LLM agent training and execution, with EnvHarness specifically addressing static environment adaptation while the advanced agentic harness covers broader orchestration patterns

Microduck: Tiny Biped Robot Both involve embodied or physical learning environments for agents — EnvHarness awakens static worlds for agent learning while Microduck is a physical biped designed for programmable agent behaviors
WikiSkill: Agent Experience into Persistent Knowledge Both frameworks focus on agent learning from experience - EnvHarness awakens static worlds for agent learning while WikiSkill persists that learned experience into transferable knowledge artifacts
Harness Engineering EnvHarness for awakening static worlds shares the core harness engineering challenge of managing environment state and progressive agent learning that this guide addresses through GC and hardening techniques

Type:	link
First published:	Wednesday, 26 August 2026 at 21:35 UTC
Location:	https://arxiv.org/abs/2608.19880
URL:	https://mattwood.fyi/i/ed53baa2

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