Today for AI
HOT RADAR
agentHEAT 0.0°

AI CURATED HOT STORY · 10/19/2026

Google Research Open-Sources Harness² for Recursive Agent Optimization

1 reports archived1 independent sourcesupdated 10/19/2026, 19:28:45
Synthesis & Latest Updates
1 Sources Cross-Validated

Google Research has open-sourced Harness², a framework for recursive agent harnessing in open-world environments. By employing dual-level recursion at both the task and system levels, it dynamically optimizes prompts, tools, and memory without altering model weights. Experiments demonstrate that Harness² significantly boosts the performance of GLM-5 and Gemini 3.5 Flash across professional benchmarks, achieving gains of up to 11.8 percentage points.

LATEST/Google Research has open-sourced Harness², a framework for recursive agent harnessing in open-world environments. By employing dual-level recursion at both the task and system levels, it dynamically optimizes prompts, tools, and memory without altering model weights. Experiments demonstrate that Harness² significantly boosts the performance of GLM-5 and Gemini 3.5 Flash across professional benchmarks, achieving gains of up to 11.8 percentage points.

HEAT TRENDHourly heat curve

3 fully observed hours
Now
0.0
Peak
15.720:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

TIMELINECoverage timeline

Total 1 reports · Latest first
  1. Hacker News AIT2·85 pts
    • Introduces 'recursive agent harnessing' using a Propose-Probe-Compose step to optimize execution without ground-truth feedback
    • Employs a dual-level architecture: task-level refinement for current execution and system-level accumulation of lessons for future guidance
    • Delivers significant performance gains for GLM-5 and Gemini on benchmarks like WorkBuddy-Bench and JobBench with a single iteration