Skip to main content
skill CX Codex

quality-refinement-autoloop — Claude Code skill

quality-refinement-autoloop turns a bad Blender result into a reusable skill fix. Diagnoses the gap, repairs either the artifact or the skill, then preps for release.

MIT 1,400 installs ★ 181 updated 2026-10-12

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add RobLe3/cc-blender-skill --skill quality-refinement-autoloop -a claude-code
Open on GitHub

What quality-refinement-autoloop does

This skill kicks in when a user rejects output: wrong geometry, broken UVs, missing animation, export failures. Instead of endless tweaks, it freezes the last accepted baseline, captures the smallest evidence set that proves the failure, and classifies the problem into eight dimensions from silhouette to orchestration.

Then it decides whether the current skill stack can fix the artifact or a generic skill needs to be added first. Repeated failures across skills lead to a harmonizer update, not a leaf-level patch. Lessons get sanitized: project names, secrets, private paths and raw logs are stripped out. Three scripts handle the loop plan, contribution sanitizing, and release readiness checks.

Hard rules block hiding geometry or UV failures behind lighting tweaks. Export support cannot be claimed if the effect only exists in Blender Python or a render sequence. No push without an explicit user request and an up-to-date remote check. Before the first Blender call the skill reads ../text-to-blender/references/mcp-backends.md and requires picking an MCP backend, official preferred for shared capabilities.

What it helps with

  • classify a Blender failure across eight dimensions and find the primary gap
  • turn a rejected run into a reusable skill instead of a one-off edit
  • strip client names, secrets and private paths before publishing a skill
  • run a sanitizer scan and skill graph audit after each change
  • stage, commit and push only on explicit request

Example prompts for the agent

  • Run the loop: the geometry on the render is wrong, check the skill and fix the artifact.
  • Write RALPH_OR_QUALITY_LOOP_REPORT.md and decide whether a new skill is needed.
  • Sanitize the skill contribution and check release readiness, but do not push without my explicit go.

How to install quality-refinement-autoloop

1Skills CLI (Vercel)

npx skills add RobLe3/cc-blender-skill --skill quality-refinement-autoloop -a claude-code

2Manually: put the files in place

~/.claude/skills/quality-refinement-autoloop/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/quality-refinement-autoloop/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about quality-refinement-autoloop

How do I install quality-refinement-autoloop in Claude Code?

Run npx skills add RobLe3/cc-blender-skill --skill quality-refinement-autoloop -a claude-code, then open ~/.claude/skills/quality-refinement-autoloop/SKILL.md.

What does quality-refinement-autoloop do?

It catches failed Blender results and runs a loop: diagnose the gap, repair the artifact or the skill, sanitize the lesson, validate. Publishing and pushing happen only when explicitly requested.

Which MCP backend does it require?

The skill asks to read ../text-to-blender/references/mcp-backends.md first and pick a backend. The official MCP is preferred for shared capabilities, and the chosen backend stays for the whole task.

Source text

SKILL.md · MIT
---
name: quality-refinement-autoloop
description: Run a self-refinement loop when a Blender result is subpar, user expectations are not met, validation fails, or repeated issues reveal missing skill depth. The loop diagnoses the gap, decides whether to repair the artifact or acquire/refine a generic skill, sanitizes lessons for reusable publication, updates docs/versioning, validates, and prepares stage/commit/push when explicitly requested.
allowed-tools: Read Bash Glob Grep mcp__blender__execute_blender_code mcp__blender__get_scene_info mcp__blender__get_object_info mcp__blender__get_objects_summary mcp__blender__get_object_detail_summary mcp__blender__render_viewport_to_path mcp__blender__search_api_docs mcp__blender_official__execute_blender_code mcp__blender_official__get_objects_summary mcp__blender_official__get_object_detail_summary mcp__blender_official__render_viewport_to_path mcp__blender_official__search_api_docs
---


## MCP backend

Use when: Subpar output, user rejects quality, repeated Blender failure, RALPH loop, skill gap diagnosis, autonomous skill refinement, sanitized skill contribution, release prep, docs/version bump, staged commit and push workflow.

Before the first Blender call, read [../text-to-blender/references/mcp-backends.md](../text-to-blender/references/mcp-backends.md). Prefer official MCP for shared capabilities; preserve the selected backend for the whole task. Tool names in recipes are client/legacy examples: map them through that reference and its agent-host guide.
# Quality Refinement Autoloop

Use this when the user says the result is wrong, ugly, not aligned, not textured, not animated, not exportable, or otherwise below expectation. The goal is not to keep tweaking blindly. The goal is to convert failure into a reusable, generic skill improvement before trying again.

## Autoloop phases

### 0. Freeze and preserve

- Stop making product changes immediately.
- Preserve the last accepted baseline and the failed artifact.
- Name the failed branch/version honestly; do not overwrite accepted outputs.

### 1. Evidence capture

Collect the smallest evidence set that proves the failure:

- user feedback quote or summary;
- source/reference files used;
- current output path/version;
- relevant render/contact sheet/overlay/audit report;
- scene/material/object inventory if the failure is inside Blender.

### 2. Diagnose failure dimension

Classify the primary gap:

- geometry / silhouette / landmarks;
- multiview/depth consistency;
- UV / atlas / texture fit;
- closed surface coverage (front/back/side);
- look/material/lighting calibration;
- animation/motion/export truth;
- orchestration/handoff between skills;
- missing validator or missing deterministic helper script.

### 3. Skill-gap decision

Ask: does the current skill stack already contain a generic method for this failure?

- If yes: run the existing skill and repair the artifact.
- If no: add/refine a generic skill first, then repair.
- If repeated failures come from skill interplay, update the harmonizer/handoff rule, not just a leaf skill.

### 4. Sanitize the lesson

Before writing a skill change:

- remove project/client/asset names;
- remove secrets, raw logs, private paths, personal data, and copyrighted source content;
- keep only reusable method, gates, scripts, and failure patterns;
- describe inputs/outputs generically;
- prefer deterministic scripts for fragile audits.

### 5. Patch skill stack

Apply the smallest publishable change:

- one concise skill or one concise section in an existing skill;
- optional helper script if the validation is repeatable;
- harmonizer update if ordering/handoff changed;
- manifest entry/version update.

### 6. Validate skill change

Required checks:

- scripts compile;
- manifest paths exist;
- skill graph audit passes with harmonizer present;
- sanitizer scan passes for project-specific terms;
- if applicable, run the new helper against the failed artifact and save a report.

### 7. Repair product only after skill gate

Rebuild or repair from the correct baseline using the improved skill. Do not reuse rejected outputs unless explicitly marked as source evidence.

### 8. Release prep when requested

Only when the user asks to publish/commit/push:

1. fetch latest remote and confirm local branch is up to date;
2. resolve/harmonize conflicts before staging;
3. bump version using semver appropriate to scope;
4. update README / plugin README / CHANGELOG / VERSIONING when relevant;
5. run sanitizer and compile/audit checks;
6. stage, commit, and push using the configured remote.

## Hard rules

- Do not bake task-specific examples into public skills.
- Do not call overlay curves/planes “texture coverage” unless the real mesh surface also passes coverage gates.
- Do not tune lighting/materials to hide geometry or UV failures.
- Do not claim export support if the effect exists only in Blender Python or a render sequence.
- Do not push without an explicit user request and an up-to-date remote check.

## Recommended artifacts

Save these near the project output when running the loop:

- `RALPH_OR_QUALITY_LOOP_REPORT.md`
- `failure_evidence/` or references to existing renders/reports
- `skill_gap_decision.json`
- `sanitization_report.json`
- `validation_report.json`

## Scripts

- `scripts/ralph_autoloop_plan.py` creates a generic failure-classification and loop plan from feedback/artifact hints.
- `scripts/sanitize_skill_contributions.py` scans skill files for project-specific or private terms before publication.
- `scripts/release_readiness_check.py` performs lightweight manifest/docs/version/path checks before commit/push.

© RobLe3 · MIT · RobLe3/cc-blender-skill

Similar skills

All skills for Game dev and 3D: Unity, Unreal, Godot, Blender →

From the same set30

The whole set →
reference-analysis-validator The reference-analysis-validator skill measures reference images and Blender renders and blocks a 1:1 claim until the metrics pass. CX Codex +1 1,500installs blender-skill-harmonizer blender-skill-harmonizer is a Claude Code skill that sets activation precedence, handoff artifacts and conflict rules when a task involves several Blender skills. CX Codex +1 1,500installs animation-quality-gate animation-quality-gate validates Blender animations with contact sheets and silhouette checks before accepting passes. Catches texture flickering and broken GLB exports. CX Codex +1 1,500installs atlas-uv-fitting atlas-uv-fitting detects texture atlas regions and fits each model part to its own UV rectangle in Blender. Helpful for fixing stretched textures and preparing clean GLB exports. CX Codex +1 1,400installs multiview-fit-loop multiview-fit-loop runs a render-compare-adjust loop until a Blender model matches supplied front, side, back and top templates across all dimensions. CX Codex +1 1,400installs wireframe-to-3d wireframe-to-3d turns orthographic wireframe PNGs into a Blender 3D model and exports it as .glb. Useful when you have front and side views but no mesh yet. CX Codex +1 1,400installs orthographic-registration orthographic-registration registers front, side, back and top orthographic references into one Blender coordinate contract. Useful when the front reads fine but side and back silhouettes drift. CX Codex +1 1,400installs contour-to-mesh contour-to-mesh builds a Blender mesh straight from a contour or mask instead of approximate primitives. Useful when the silhouette has to match a front template exactly. CX Codex +1 1,400installs

Project sponsors

Companies that keep this analytics open