Image Analysis for Visual Quality Assessment

VerifiedSafe

Delegates image analysis to GLM-4.6v subagent for visual quality assessment of game screenshots and sprites. It checks for issues like blocky backgrounds, transparency problems, and style consistency against Harvest Moon SNES references. Use for batch analysis or when extensive visual QA is needed to save tokens.

Sby Skills Guide Bot
DevelopmentIntermediate
6/2/2026
Claude Code
#image-analysis#game-development#pixel-art#quality-assessment#sprite-validation

Recommended for

Automated summary

This skill delegates image analysis to a GLM-4.6v subagent for visual quality assessment of game screenshots and sprites, following Harvest Moon SNES style.

Strengths

  • Specialized prompts for pixel art and stylistic consistency
  • Batch processing via parallel subagents to save tokens
  • Concrete actionable feedback on transparency, proportions, and colors
  • Direct Claude integration for small tasks without overhead

Limitations

  • Limited to game analyses and specific visual style
  • Requires access to GLM-4.6v subagent via MCP
  • Not suitable for general image or photographic analysis
When to use it

Use this skill to validate game assets, sprites, or screenshots against precise pixel art quality criteria.

When not to use it

Do not use it for non-game images or when the analysis can be done directly by Claude without delegation.

Security analysis

Safe
Quality score85/100

The skill provides instructions for image quality analysis using an MCP tool and example shell commands. None of the commands are destructive or exfiltrating; they are limited to benign image generation, processing, and version control status checks.

No concerns found

How this analysis is made

Examples

Analyze Screenshot
Analyze this game screenshot for visual quality. Check for: blocky backgrounds, tiling seams, sprite transparency issues, pixel art scaling problems, overall Harvest Moon SNES readability. Provide specific, actionable feedback.
Analyze Sprite Asset
Analyze this pixel art sprite for Harvest Moon SNES style quality. Check: transparency (clean edges), silhouette (strong/readable), proportions (compact not stretched), color palette (dusk-friendly), pixel detail (minimal noise). Reference: assets/sprites/PLACEHOLDER_ASSET_SPEC.txt
Transparency Check
Check this sprite for transparency issues: - Halo effect (light/dark ring around edges) - Color bleeding (background showing through) - Harsh edges (no smooth curves on diagonals) - Partial transparency (should be opaque or transparent, not in-between). Report specific pixel locations and severity.

name: image-analysis description: GLM-4.6v image analysis subagent for game screenshot and sprite quality assessment

Image Analysis Skill

Purpose: Delegate image analysis tasks to GLM-4.6v subagent for visual quality assessment of game screenshots and sprites.

When to Use

Invoke this skill when:

  • New sprites are generated (via glm-image-gen)
  • Screenshots captured during playtesting
  • Visual quality validation needed before commits
  • Checking for blocky backgrounds or transparency issues
  • Comparing assets against Harvest Moon SNES style

Direct vs Delegated Analysis

Claude Opus can analyze images directly via the Read tool. Choose the right approach:

| Scenario | Approach | Why | |----------|----------|-----| | 1-3 images | Claude Direct | No overhead, faster | | Quick quality check | Claude Direct | Immediate feedback | | Batch analysis (10+) | MiniMax Parallel | Token efficiency | | Extensive visual QA | MiniMax Parallel | 50x cheaper |

Claude Direct Pattern

Read image file → Analyze in-context → Continue workflow

MiniMax/GLM Delegation Pattern (for batches)

Spawn parallel agents → Each analyzes subset → Aggregate results

Default: Use Claude Direct for small tasks, delegate batches to save tokens.

Quick Start

Analyze Screenshot

mcp__4_5v_mcp__analyze_image \
  --imageSource "temp/screenshots/[filename].png" \
  --prompt "Analyze this game screenshot for visual quality. Check for: blocky backgrounds, tiling seams, sprite transparency issues, pixel art scaling problems, overall Harvest Moon SNES readability. Provide specific, actionable feedback."

Analyze Sprite Asset

mcp__4_5v_mcp__analyze_image \
  --imageSource "assets/sprites/placeholders/[filename].png" \
  --prompt "Analyze this pixel art sprite for Harvest Moon SNES style quality. Check: transparency (clean edges), silhouette (strong/readable), proportions (compact not stretched), color palette (dusk-friendly), pixel detail (minimal noise). Reference: assets/sprites/PLACEHOLDER_ASSET_SPEC.txt"

Analyze NPC Sprite

mcp__4_5v_mcp__analyze_image \
  --imageSource "assets/sprites/placeholders/npc_[name].png" \
  --prompt "Analyze this 64x64 NPC sprite. Target proportions: ~48h x 32w centered. Check: strong silhouette, Mediterranean theme alignment, clean transparency, no elongation. Harvest Moon SNES style reference."

Standard Prompts

Full Quality Assessment

Analyze this image for Circe's Garden game quality standards:

Visual Style:
- Harvest Moon SNES readability (clean pixel art, strong silhouettes)
- Stardew Valley influence (but more mythic/moody)
- Mediterranean ancient Greek island theme
- Dusk-friendly palette (muted, not oversaturated)
- 32px tiles, 2x camera zoom

Check For:
1. Background issues: blocky patterns, tiling seams, color banding
2. Transparency: clean edges (no halos), proper alpha channel
3. Pixel art: minimal noise, clear details, proper scaling
4. Proportions: compact bodies, not stretched/elongated
5. Colors: natural earth tones, muted jewel accents
6. Lighting: soft ambient, no harsh specular
7. Silhouette: strong, recognizable at small size

Provide specific feedback: what works, what needs fixing, file references.

Transparency Check (Focused)

Check this sprite for transparency issues:
- Halo effect (light/dark ring around edges)
- Color bleeding (background showing through)
- Harsh edges (no smooth curves on diagonals)
- Partial transparency (should be opaque or transparent, not in-between)

Report specific pixel locations and severity.

Style Comparison

Compare this asset to Harvest Moon SNES style:
- Clean pixel art with minimal noise
- Strong silhouette (recognizable at 32x32)
- Soft, natural color palette
- Proper alpha blending
- No anti-aliasing on edges
- Cozy, approachable aesthetic

Report what matches style and what deviates.

Output Format

Expected response should include:

## Analysis Summary
[One-line assessment: PASS / NEEDS WORK / FAIL]

### Issues Found
**Critical (P0):**
- [Specific issue with location]

**Important (P1):**
- [Issue description]

**Minor (P2):**
- [Minor detail]

### What Works
- [Positive aspects to preserve]

### Recommendations
1. [Actionable fix 1]
2. [Actionable fix 2]

Common Issues

| Issue | Detection | Fix | |-------|-----------|-----| | Blocky background | Visible pixel grid, tiling patterns | Regenerate with better seam handling | | Halo effect | Light/dark ring around sprite | Fix alpha channel, clean edges | | Color bleeding | Background color in transparent pixels | Re-export with proper transparency | | Stretched proportions | Elongated body/limbs | Regenerate with compact proportions | | Oversaturated colors | Too bright/vibrant | Adjust palette to dusk-friendly tones | | Poor silhouette | Not recognizable at small size | Strengthen outline, simplify details |

File Locations

Input:

  • Screenshots: temp/screenshots/
  • Sprites: assets/sprites/
  • Placeholders: assets/sprites/placeholders/

Reference:

  • Asset specs: assets/sprites/PLACEHOLDER_ASSET_SPEC.txt
  • Style guide: assets/sprites/PLACEHOLDER_README.md

Integration

With glm-image-gen

# After generating new sprite
./generate-image.sh '{"subject":"..."}' 1024x1024 standard
./process-image.sh "$TEMP" "32x32" "assets/sprites/output.png"

# Then analyze quality
mcp__4_5v_mcp__analyze_image --imageSource "assets/sprites/output.png" --prompt "[standard prompt]"

With HPV (Playtesting)

# During screenshot review
ls temp/screenshots/
mcp__4_5v_mcp__analyze_image --imageSource "temp/screenshots/world_map.png" --prompt "[screenshot prompt]"

Pre-commit Check

# Before committing asset changes
git status
mcp__4_5v_mcp__analyze_image --imageSource "assets/sprites/new_asset.png" --prompt "[validation prompt]"

Subagent Role

Full subagent configuration: .claude/roles/image-analysis-subagent.md

Model: GLM-4.6v (vision-capable) Tool: mcp__4_5v_mcp__analyze_image Purpose: Specialized visual quality assessment

Success Criteria

Analysis is effective when:

  • Issues match human findings (80%+ accuracy)
  • False positives < 20%
  • Actionable feedback provided
  • Response time < 30 seconds
  • Consistent report format

Related:

  • .claude/skills/glm-image-gen/SKILL.md (sprite generation)
  • assets/sprites/PLACEHOLDER_ASSET_SPEC.txt (asset specifications)
  • .claude/roles/image-analysis-subagent.md (full subagent docs)

[Kimi Code CLI - 2026-01-26]

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/image-analysis-for-visual-quality-assessment-mm5t8a8m/download' --output '.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m/SKILL.md.part' && mv '.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m/SKILL.md.part' '.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m/SKILL.md' && test -s '.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/image-analysis' && cp -R '.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m' '.claude/skills/image-analysis'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m'

openskills read 'image-analysis-for-visual-quality-assessment-mm5t8a8m'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/image-analysis-for-visual-quality-assessment-mm5t8a8m/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills