Ablation study: domain-scoped editing worth up to 3.7 points in HARNESS² agent framework
morgymcg · x · 2026-10-20
The author shares an ablation excerpt from the HARNESS² paper, noting the domain-agnostic vs domain-scoped split may matter a lot. On WorkBuddy-Bench and JobBench (Gemini 3.5 Flash executor): removing the improvement library costs 3.1/1.8 points, removing scoping costs 2.9/3.7, and removing composition costs 1.6/1.9 — with domain-scoped candidate editing the biggest single contributor.
More from coding & agent
- Bots that should be skills: auditing a Grok Bot fleet with six tests — bfrench · 2026-10-20
- Dev's code review setup: agent diagrams before/after flows on a canvas for Q&A — max__drake · 2026-10-20
- tsrs: an agent-written Rust port of TypeScript is 8x faster than TS 7.0 with 70% less memory — cnakazawa · 2026-10-20
- ASCII Magic launches 634-piece MIT-licensed animated ASCII art library you can wire into coding agents — sull · 2026-10-20
- Dev details how to let an MCP agent delete files safely: HMAC preview tokens and deny lists — qupit · 2026-10-20
- Cloudflare adopts tsrs, a Rust port of TypeScript that is 8x faster with 70% less memory — cnakazawa · 2026-10-20