Harness²
AIGoogle Research's open-source framework for recursive agent harnessing that refines an agent's prompts, tools and memory per task.
Harness²
Google Research's open-source framework for recursive agent harnessing that refines an agent's prompts, tools and memory per task.
About Harness²
Harness² is a plug-and-play framework for recursive agent harnessing in an open world, where agents face tasks unseen during harness design. A harness is the system around a model, including prompts, tools, memory, skills and orchestration, and Harness² optimizes it while keeping model weights fixed. It works at two levels: task-level recursion refines the execution harness through a contrastive Propose-Probe-Compose step, and harnessing-level recursion updates a persistent improvement harness with lessons from each task. The release supports parallel and sequential recursion, OpenCode and Codex executors, and the JobBench, LAB and WorkBuddy Bench benchmarks. It is released under Apache 2.0.
Key Features
Use Cases
Frequently asked questions about Harness²
What is Harness²?
Google Research's open-source framework for recursive agent harnessing that refines an agent's prompts, tools and memory per task.
How does Harness² work?
Harness² works by combining Task-Level Recursion: A contrastive Propose-Probe-Compose step proposes edits, probes their effects on the same task and composes a refined harness without ground-truth feedback., Harnessing-Level Recursion: A persistent improvement harness is updated with adopted harnesses and evidence-supported lessons from each task., Eight Editable Components: Exposes system prompt, guardrails, memory, tool guidance, subagents, skills, scripts and plugins as tunable parts., Parallel and Sequential Modes: Scale the task-level recursion budget by exploring candidates in parallel or refining sequentially., Multiple Executors: Run refinement with OpenCode and Codex executors. to help users with Agent Harness Research: Study how contrastive harness edits affect agent performance across benchmarks., Improving Base Agents: Raise results of models such as GLM 5 and Gemini 3.5 Flash over their default harnesses., Professional Task Evaluation: Evaluate agents on legal, office and coding tasks using bundled benchmark guides., Tuning Coding Agents: Refine system prompts, skills and subagents for OpenCode or Codex executors., Reproducible Experiments: Run, judge and report benchmark experiments with resume support..
What are the main features of Harness²?
Key features include Task-Level Recursion: A contrastive Propose-Probe-Compose step proposes edits, probes their effects on the same task and composes a refined harness without ground-truth feedback., Harnessing-Level Recursion: A persistent improvement harness is updated with adopted harnesses and evidence-supported lessons from each task., Eight Editable Components: Exposes system prompt, guardrails, memory, tool guidance, subagents, skills, scripts and plugins as tunable parts., Parallel and Sequential Modes: Scale the task-level recursion budget by exploring candidates in parallel or refining sequentially., Multiple Executors: Run refinement with OpenCode and Codex executors..
Who is Harness² for?
Harness² is useful for anyone interested in Agent Harness Research: Study how contrastive harness edits affect agent performance across benchmarks., Improving Base Agents: Raise results of models such as GLM 5 and Gemini 3.5 Flash over their default harnesses., Professional Task Evaluation: Evaluate agents on legal, office and coding tasks using bundled benchmark guides., Tuning Coding Agents: Refine system prompts, skills and subagents for OpenCode or Codex executors., Reproducible Experiments: Run, judge and report benchmark experiments with resume support..
How much does Harness² cost?
Harness² is free to use.
How do I get started with Harness²?
Visit https://github.com/google-research/harness2 to sign up and explore Harness².
Explore more AI Ai Agents tools
Browse all Ai Agents tools →Browse by use case: Automation & Productivity
Compare Harness²: vs Scout by Fini · vs Ruflo · vs The New Beings · vs Kai Resolve by Gleap