skills.rest
AvatarGanymede
AvatarGanymedeCOMMUNITY
Explore all skills by this author →
GitHub

gui-improve

Iteratively fixes merged CRITICAL review findings and reruns dual-lane GUI verification gates.

Updated Jun 16, 2026
Search Tags:#code review#pipeline automation#mvvm#critical issues#iterative fixing#unity gui#bias guard
One-click install
npx skills add https://github.com/AvatarGanymede/dev-gui-plugin --skill gui-improve-avatarganymede
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: gui-improve
Source: https://github.com/AvatarGanymede/dev-gui-plugin/tree/main/skills/gui-improve
Command: npx skills add https://github.com/AvatarGanymede/dev-gui-plugin --skill gui-improve-avatarganymede

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? After a GUI code review reports CRITICAL issues (Type-B logic findings plus Type-A compile/syntax failures), someone must fix them and re-verify without letting the fixer grade their own work. This Skill automates that bounded fix-and-re-review loop for Unity MVVM GUI code. ## Core Features & Use Cases - Bounded Iteration Loop: Fixes merged CRITICAL and MAJOR issues from GUI_REVIEW.md and GUI_VERDICT.json, then reruns the review, for a maximum of 2 rounds. - Bias Guard Re-Review: Spawns a fresh gui-reviewer subagent each round with no prior results, fix claims, or thread IDs, so fixes are judged independently. - Escalation Without Blocking: Unresolved CRITICALs after 2 rounds are written to HUMAN_REVIEW.md with file:line references, and the pipeline continues to the learning phase. - Use Case: A BagPanel review reports a null reference and missing event unsubscription as CRITICAL; this Skill fixes both, re-runs the dual-lane review, and logs the score improvement in GUI_IMPROVEMENT_LOG.md. ## Quick Start Ask the pipeline to fix the merged CRITICAL issues from the latest gui-review verdict for the current panel and rerun the review gates.

Frequently Asked Questions about gui-improve

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix CRITICAL issues found in a GUI code review?▼

Read the merged CRITICAL set from GUI_REVIEW.md and GUI_VERDICT.json, fix issues by priority (CRITICAL then MAJOR, including compile and syntax errors), then rerun the review. This loop repeats for at most 2 rounds.

What is a Bias Guard in automated code review?▼

Bias Guard means each re-review spawns a completely fresh reviewer subagent that never sees prior review results, fix claims, or previous thread IDs. It only receives current code, the plan, and shared references, so the fixer cannot influence the verdict.

What happens if CRITICAL issues remain after the fix rounds?▼

After 2 rounds, any unresolved CRITICAL items are written to HUMAN_REVIEW.md with file:line references and reasons. The pipeline does not pause; it continues to the learning phase and leaves final resolution to human review.

When should the gui-improve iteration loop be triggered?▼

Trigger it only when the review phase reports merged CRITICAL findings, meaning Type-B reviewer CRITICALs or Type-A mechanical gate failures such as compile or Lua syntax errors. It is not intended for clean reviews.

Can I manually invoke gui-improve from the command menu?▼

No. The Skill sets user-invocable to false, so it does not appear in the slash menu. It runs automatically inside the pipeline, typically via the run command or after gui-review reports CRITICALs.

Related Skills

Explore More Software Engineering Skills→
igot-aiigot-ai

Add UGUI View

Create Unity UGUI screens with Model-View bindings and prefab generation.

Official
Advanced
HerxinsasaHerxinsasa
1

problem-summary

Structure troubleshooting outcomes into standardized 7-section knowledge base entries.

Community
Intermediate
gqf2008gqf2008

requesting-code-review

Automate pre-commit verification with static security scans and quality gates.

Community
Intermediate
ivegamsftivegamsft

human-in-the-loop

Set up confirmation gates, escalation triggers, and approval workflows for automated tasks.

Community
Advanced