# QoderAI/better-harness

An open-source Harness Engineering platform for coding agents—define harnesses as code, run controlled experiments, inspect evidence, and compare outcomes.

Repository: https://github.com/QoderAI/better-harness
Canonical: https://ross.abutalabs.com/products/better-harness
Homepage: https://qoderai.github.io/better-harness/
Language: JavaScript
License: MIT
License Family: permissive
Topics: agent-plugin, claude-code, codex, cursor, harness-design, harness-engineering, loop-engineering, qoder, skill
Last push: 2026-08-26T16:48:56+00:00

## Health v2 (maintenance only)
Score: 68/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 64, longevity 3
- inputs: {"age_days": 43, "days_push": 7, "days_rel": 29, "gap_med": null, "n_releases_24m": 1}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1976, forks 165 (observed 2026-08-28T04:06:01.277715+00:00)

## What it is
Better Harness is an open-source Harness Engineering platform for coding agents that lets you define harnesses as code, run controlled experiments, and inspect evidence. It integrates as a plugin or extension across hosts like Claude Code, Codex, Cursor, and Qoder, turning project and session evidence into prioritized improvements with verifiable next steps.

## Use cases
- improve my coding agent workflow loop
- analyze claude code sessions for workflow improvements
- run controlled experiments on coding agent harnesses
- generate evidence-based reports of agent coding gaps
- compare outcomes across coding agent runs
- define agent harnesses as code
- get prioritized fixes with acceptance checks from agent sessions

## When to choose
- You already use a supported coding agent host (Claude Code, Codex, Cursor, Qoder, Qwen Code, GitHub Copilot CLI) and want to improve the loop around it
- You want evidence-bounded, reviewable reports with impact, scoped repairs, and acceptance checks
- You need missing or partial evidence to stay explicit rather than guessed

## When to avoid
- You need a general-purpose agent framework for building agents from scratch
- You use a coding agent host without an adapter (check the Host Adapter Matrix first)
- You want fully automated fixes without human review of evidence

## Facets
- artifact type: plugin
- maturity: active
- function: agent-framework, developer-tools, analytics, workflow-automation, testing
- domain: developer-tools
- platform: cli, editor-plugin, cross-platform
- tags: harness-engineering, coding-agents, claude-code, codex, cursor, agent-work-loop, evidence-reports, loop-engineering, ai-agents, automation, nodejs

## Member repositories
- QoderAI/better-harness (main) score 68

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:01.277715+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:04:48.937282+00:00, confidence not recorded.
  - readme: https://github.com/QoderAI/better-harness (fetched 2026-08-28T04:06:01.277715+00:00, sha 4e5ea44ce827)
  - homepage: https://qoderai.github.io/better-harness/ (fetched 2026-08-29T10:44:10.762883+00:00, sha 7ee962caf568)
- Data as of 2026-08-30T08:39:29.467469+00:00.