Back to articles
DevPickView feed
Hacker NewsHN submitter: matt_d
Original published:

HarnessTax: How Much Does the Harness Matter for Coding Agents?

AI summary
Editorial policy

コーディングエージェントのパフォーマンスにおいて、モデル自体よりも周辺ツールや実行環境（ハーネス）の影響が非常に大きいことを定量的に分析した研究結果が公開された

Key points

  • •同じLLMモデルであっても、使用する開発ツールチェーンやエージェントの制御ロジック（ハーネス）によって成功率が大きく変動する
  • •研究では、最適化されたハーネスを使用することで、モデルの能力を最大限に引き出し、タスク遂行率を大幅に向上させられることを示した
  • •AIエージェントの実務導入においては、モデル選定だけでなく、周囲のインフラやプロンプト設計の最適化が成功の鍵となる
#AI Agents#Research & Papers
Read the original(opens in a new tab)

Related articles(AI Agents)

See more
TechCrunch

Replit is no longer working with Anthropic, CEO says

Replit CEO and co-founder Amjad Masad stated at TechCrunch Disrupt 2026 that the company is no longer working with Anthropic.

Hacker News

Why harnesses matter for energy (and token) efficiency

Analysis of a month of agent traffic reveals that the harness, not just the model, determines the energy consumption of coding sessions.

The Verge

Tesla driver says he was asleep and using FSD when he crashed into highway patrol cars

The California Highway Patrol warned that Full Self-Driving is not fully autonomous after a Tesla driver admitted to being asleep while using the feature during a crash into patrol cars.

Hacker News

How the Claude Code team does code review

The Claude Code team optimizes for the minimum amount of human review time that does not cause reverts or incidents, addressing the volume of code produced by agents.

Google Research

Grounding AI agents in biological reality: Building blocks for autonomous wet labs

Google Research published a piece on building blocks for autonomous wet labs that ground AI agents in biological reality.