Harness² vs openrig: Features, Pricing & Which Is Better (2026)
A side-by-side comparison of Harness² and openrig — features, pricing, and ideal use cases — to help you decide which AI tool fits your workflow.
H
Harness²
Google Research
Google Research's open-source framework for recursive agent harnessing that refines an agent's prompts, tools and memory per task.
Key features
- 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.
- Built-In Benchmarks: Setup and run commands for JobBench, LAB (legal work) and WorkBuddy Bench (office, web and code).
- Credential-Free Demo: A local demo with deterministic editors runs both recursion modes without model credentials.
Best for
- 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.
o
openrig
mvschwarz
OpenRig is open-source software to define a team of Claude Code and Codex agents in YAML and run them as one persistent, managed system.
Key features
- YAML Rig Definitions: Describe an agent team with roles in YAML and boot it with a single command.
- Mixed Harnesses: Run Claude Code and Codex agents together in the same rig, managed as one system.
- Lead Agent Coordination: Talk to a lead agent that coordinates specialists across teams and reports back decisions needing attention.
- Persistent Teams: Keep team work and context at stable addresses instead of a pile of terminal sessions.
- Terminal UI: View the rig as a graph or a table of seats showing runtime, model, context and state, with per-seat detail.
- Guided First Use: A getting-started path installs, launches a two-agent team in your repository and produces one reviewed change.
- Permission Allowances: Optionally lets your agent add scoped native rules so OpenRig commands run without repeated prompts, with an undo.
- CLI Install: Distributed as @openrig/cli on npm, requiring Node.js 22 or 24 and tmux on macOS or Linux.
Best for
- Multi-Agent Coding Teams: Stand up a persistent team of coding agents with defined roles for a repository.
- Cross-Harness Workflows: Combine Claude Code and Codex strengths within one coordinated team.
- Outcome-Driven Delegation: Describe a desired outcome to a lead agent and let it split work among specialists.
- Agent Monitoring: Watch seats, models, context and state of many agents from one terminal UI.
- Reviewed Changes: Have agents produce a change in a repo and another review it before you merge.
- Agent Research Experiments: Prototype and study networks of cooperating agents on your own machine.