Use when turning repeated failures, weak evals, weak handoffs, or trace blind spots into concrete improvements to the development harness.
Scanned 9/11/2026
Install to Claude Code
npx -y skills add ponybunny-ai/ponybunny --skill harness-improvement-loop --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Harness Improvement Loop?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/ponybunny-ai-harness-improvement-loop)More formats (shields.io, HTML) on the badges page.
---
name: harness-improvement-loop
description: Use when turning repeated failures, weak evals, weak handoffs, or trace blind spots into concrete improvements to the development harness.
---
# Harness Improvement Loop
## Use this skill when
- the same class of issue keeps recurring
- generated work often needs the same correction
- verification is repeatedly weak
- handoffs are not good enough
- traces are insufficient for diagnosis
## Goals
- improve the harness instead of only fixing local artefacts
- reduce repeated waste
- strengthen planning, generation, evaluation, and handoff quality
## Process
1. Collect evidence from failures, traces, or weak reviews.
2. Classify the recurring pattern.
3. Identify the likely harness-level cause.
4. Propose a small high-leverage improvement.
5. Define rollout steps.
6. Define how to validate the improvement.
## Output contract
Return:
- recurring pattern
- evidence
- likely harness cause
- improvement proposal
- rollout plan
- success criteria
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!