Sign InOpen Brain
arXivPaperNeeds Review

An Empirical Study of Harness Design for Coding Agents

Harness components pay off differently by model and budget. Elide before summarizing, use planning selectively, and avoid elaborate tools when the model is already strong with bash.

arXiv · Sep 17, 2026
Open Source Open MarkdownOpen JSON
Source Summary

Researchers tested **176 matched settings** across **four models**, varying planning, action space, and context management on SWE-Bench Verified and Terminal-Bench 2.1. Context handling mattered more as budgets tightened, chiefly by preventing overflow.

Practical Implication

Stage rule-based elision before LLM summarization. Use planning as an accuracy scaffold for weaker models and a cost control for stronger ones; offer predefined tools when bash skill is weak, but consider bash-only operation for capable models on CLI-heavy work.

Agent-Ready Context
Researchers tested **176 matched settings** across **four models**, varying planning, action space, and context management on SWE-Bench Verified and Terminal-Bench 2.1. Context handling mattered more as budgets tightened, chiefly by preventing overflow.

Stage rule-based elision before LLM summarization. Use planning as an accuracy scaffold for weaker models and a cost control for stronger ones; offer predefined tools when bash skill is weak, but consider bash-only operation for capable models on CLI-heavy work.

Recoverable elision added machinery without an accuracy gain because models rarely used recovery. The evidence spans **five context strategies** and **four window budgets**, but the abstract provides no effect sizes and covers only two benchmarks.
Connected Context · Feed7 Judgment

This supplies comparative evidence for choices that prior harness guidance treated mainly as design principles. It narrows the highest-leverage coding-agent controls to budget-aware context management, model-dependent planning, and action spaces matched to tool competence, while warning that extra recovery machinery is not automatically useful. Its two-benchmark scope does not establish the broader production controls needed for stateful or irreversible work.

Harness Engineering: Building the Production Cage for Powerful Domain Agents — Mike Chambers, AWSThe production-harness framework defines the surrounding system broadly; this study adds matched empirical evidence about which planning, context, and tool-interface choices matter within that boundary.Agents Without Code: Skills, YAML, and Filesystems Replaced Python — Philipp Schmid, Google DeepMindThe reduced-orchestration approach favors general tools and instructions, while these results qualify that choice by showing that predefined tools can still help models with weaker bash competence.AI Agents Are Just Distributed Systems Now — Salman Munaf, TikTokThe study measures accuracy and context behavior on coding benchmarks, whereas this candidate identifies external-state semantics and recovery controls that remain necessary beyond benchmark task completion.
Context Map
agentcoding#harness-engineering#context-engineering#tool-use
Uncertainty
Recoverable elision added machinery without an accuracy gain because models rarely used recovery. The evidence spans **five context strategies** and **four window budgets**, but the abstract provides no effect sizes and covers only two benchmarks.