Microsoft's Self-HPO Lets Coding Agents Rewrite Their Own Harness

Microsoft's Self-HPO uses reinforcement learning to alternately optimize a coding agent's harness and policy, letting the agent redesign its own runtime environment based on failed attempts. The approach boosts success rates by 7 percentage points.

2026-10-20 ~ 2026-10-21 · 2 related posts

• Microsoft Research proposes Self-HPO for self-evolving coding agents — MicrosoftResearch · 2026-10-20
• Microsoft's Self-HPO: Coding Agent Redesigns Its Own Harness, +7 Points on SWE-bench — mark_k · 2026-10-21