PixelMagic-PhotoLogic-9z

Professional image post-processing tool. Supports multiple editing styles (apocalyptic cinematic, Japanese fresh, vintage film, etc.), auto-detects and insta...

Security Scan

The skill's requests and instructions are internally consistent with an image-postprocessing tool: it manipulates local image files with ImageMagick, creates isolated work directories, and uses multimodal checks for validation; no unrelated credentials, installs from unknown URLs, or hidden endpoints are requested.

Purpose & Capability

Name/description match the runtime instructions: the SKILL.md focuses on ImageMagick-based editing, RAW support, batch processing, presets, and file management. The _skillhub_meta.json also lists ImageMagick as a system dependency — coherent with the stated purpose.

Instruction Scope

Instructions are specific to local image processing (create workspace, generate intermediate JPGs, record per-step params, build a final magick command). Two notes: SKILL.md instructs the agent to perform multimodal validation of intermediate JPGs (which may cause images to be sent to whatever multimodal model/endpoints the platform uses), and it says to auto-detect and prompt to install ImageMagick but gives no concrete install steps — both are expected for this skill but are privacy/operational considerations rather than incoherent behavior.

Install Mechanism

This is instruction-only with no install spec or downloaded code; that reduces disk-write risk. The only install-related claim is auto-detect/install ImageMagick — there is no packaged installer URL or extract step in the skill itself, so nothing unexpected is being pulled by the skill bundle.

Credentials

The skill requests no environment variables, credentials, or external config paths. All file and directory actions are local and directly related to image processing; there is no disproportionate access requested.

Persistence & Privilege

Flags are default (always:false, user-invocable:true). The skill does not request persistent/always-on presence or modify other skills. It writes workspace files and params for its own operation only — expected for this functionality.

Assessment

This skill appears to do what it says: local ImageMagick-based iterative image editing with isolated work folders and recorded parameters. Before using it, consider: (1) ImageMagick installation — the skill may prompt the agent/platform to run package-manager commands to install it, so confirm you trust those operations; (2) privacy — the multimodal 'validation' step implies the platform's multimodal model will view intermediate JPGs (which may be uploaded to remote services); avoid processing sensitive images unless you know where multimodal inference runs; (3) filesystem effects — the skill creates workspace directories and writes intermediate JPGs and params (review and clean them if needed); and (4) if you want to audit execution, ask for the exact shell/installation commands the agent will run (the SKILL.md provides magick examples but not install commands). If those points are acceptable, the skill is coherent and proportionate to its stated purpose.

Professional Image Post-Processing Tool

Features

• ✅ Auto-detect and install ImageMagick
• ✅ Multiple editing style presets
• ✅ Support for RAW/JPG/PNG/TIFF formats
• ✅ Batch processing support
• ✅ Custom preset saving
• ✅ Isolated intermediate file management

Editing Style Presets

Style	Characteristics	Use Cases
apocalyptic	Low saturation, teal-orange tones, vignette, grain	Post-apocalyptic, sci-fi, cinematic
japanese	High brightness, low contrast, cyan-green tint	Japanese fresh style, daily records
vintage	Warm tones, faded look, soft focus	Retro, nostalgic
bw-high	Black & white, high contrast	Documentary, artistic
custom	User-defined parameters	Personalized needs

Usage

Single Image Processing

from scripts.editor import ImageEditor

editor = ImageEditor()
result = editor.process(
    input_path="photo.dng",
    style="apocalyptic",
    output_name="my_photo"
)

Batch Processing

editor.batch_process(
    input_dir="./photos",
    style="japanese",
    output_dir="./output"
)

Custom Presets

# Create custom preset
editor.create_preset(
    name="my_style",
    params={
        "brightness_contrast": "-10x30",
        "modulate": "100,70",
        "tint": "#3a5a4a",
        "gamma": 0.95,
        "sharpen": "0x1.2",
        "noise": 0.3,
        "vignette": True
    }
)

Workflow (Iterative Step-by-Step Editing)

Core Principles

• Must use iterative step-by-step editing to ensure each step meets quality standards before proceeding, ultimately generating the final image from the original without compression loss.
• Do not increase contrast unless the style specifically calls for high contrast.
• Unless style dictates otherwise, preserve smooth tonal transitions and retain highlight and shadow details.
• When reviewing intermediate and final images with multimodal models, watch for artifacts like color banding, noise, or color patches caused by excessive parameter adjustments.

Detailed Steps

1. Initialization

• Detect ImageMagick, prompt for installation if not found
• Analyze original image (resolution, format, color space)
• Create isolated working directory based on original filename (to avoid file conflicts between different images; append -001, -002... for duplicates):

  workspace/image-editor-work/
  ├── {filename}/         # Directory named after original file
  │   ├── temp/           # Intermediate processing files
  │   │   ├── step01.jpg
  │   │   ├── step02.jpg
  │   │   └── params/     # Successful parameter records per step
  │   │       ├── step01_params.txt
  │   │       ├── step02_params.txt
  │   │       └── ...
  │   └── final/          # Final output
  │       └── xxx_final.jpg
  ├── {filename2}/        # Separate directory for another image
  └── ...
  

• Example: When processing IMG_20260329_111232.dng, create image-editor-work/IMG_20260329_111232/ directory

2. Iterative Step-by-Step Editing (Key Process)

Core Principle: Preserve original dynamic range and bit depth

For all images (whether RAW or other formats), each step must start from the original file, stacking previously verified parameters to avoid dynamic range and bit depth loss from intermediate JPGs.

Step A - Generate Intermediate JPG (For validation only, not part of editing chain)

Core Principle: Always start from original file

Each processing step starts from the original file, carrying all previously verified parameters + new parameters for the current step, generating an intermediate JPG for multimodal validation only.

• Command format examples:

  # Step 1: Start from original
  magick original.dng -brightness-contrast -5x20 -quality 100 step01.jpg
  
  # Step 2: Still from original, stack Step 1 params + new params
  magick original.dng -brightness-contrast -5x20 -modulate 110,80 -quality 100 step02.jpg
  
  # Step 3: Continue from original, stack Step 1+2 params + new params
  magick original.dng -brightness-contrast -5x20 -modulate 110,80 -fill "#3a5a6a" -tint 15 -quality 100 step03.jpg
  

• JPG quality must be highest: -quality 100

• Strictly prohibited: -resize or -crop: Do not change original resolution unless user explicitly requests

• Note: ImageMagick 7+ uses magick command, no need for magick convert

Important: Intermediate JPGs are for multimodal model validation only, never used as input for next step.

Step B - Multimodal Validation

• Use multimodal model to review generated intermediate JPG
• Determine if editing goal for this step is achieved
• If goal NOT achieved:
  • Analyze cause, adjust magick parameters
  • Restart from original file with all previously verified params + adjusted new params
  • Repeat this step until达标 (standard met)
• If goal achieved:
  • Write the new magick parameters for this step to temp/params/stepXX_params.txt
  • Continue from original file for next step with all verified params (including just recorded)

Key Memory:

• Each step re-reads original file
• Parameters accumulate: Step N = original + param1 + param2 + ... + paramN
• Intermediate JPGs are view-only, next step still starts from original

3. Loop Execution

• Repeat "Step A → Step B" process for next editing step
• Record parameters to corresponding stepXX_params.txt after each step达标
• Continue until all editing steps complete

4. Generate Final Image (Critical Step)

Strictly prohibited: Use last intermediate JPG as final output.

Correct approach:

1. Read original image
2. Read each step's parameter file stepXX_params.txt in sequence
3. Build complete magick command chain, concatenating all parameters in editing order
4. Generate final image from original in one pass:

   magick original.dng \
     -brightness-contrast -5x20 \
     -modulate 110,65 -fill "#2a5a6a" -tint 20 -gamma 0.95 \
     -size 4096x3072 radial-gradient:black-white -compose multiply -composite \
     -sharpen 0x1.5 \
     -attenuate 0.5 +noise gaussian \
     -quality 100 final_xxx.jpg
   

Why do this?

• Intermediate JPGs suffer JPEG compression loss from multiple saves
• Applying all parameters from original in one pass avoids layer-by-layer compression
• Final image has highest quality

5. Wrap-up & Delivery

File Saving:

• Copy final image to final/ directory
• Optional: Clean up intermediate files or keep for review
• Record processing log

Deliver to User:

1. Final file path (must provide)

   Editing complete! Final image saved to: {final_path}
   

2. Send image to user

Important: Regardless of open_result_view success, always explicitly inform user of local file path to ensure they can find the final product.

Workflow Diagram

Unified Processing Flow (all formats):

Original
   │
   ├──► Step 1 params ──► Generate intermediate JPG ──► [Multimodal check] ──► Standard met?
   │                                                                         │
   │                                                                        No ──► Adjust params ──► Regenerate from original
   │                                                                         │
   │                                                                       Yes ──► Record params to step01_params.txt
   │
   ├──► Step 1+2 params ──► Generate intermediate JPG ──► [Multimodal check] ──► Standard met?
   │                                                                           │
   │                                                                          No ──► Adjust params ──► Regenerate from original
   │                                                                           │
   │                                                                         Yes ──► Record params to step02_params.txt
   │
   ├──► Step 1+2+3 params ──► Generate intermediate JPG ──► [Multimodal check] ──► Standard met?
   │                                                                             │
   │                                                                            ...
   ▼
[Final Generation] ──► Apply all params from original in one pass ──► Final image (highest quality)

Key Points:

• Each step starts from original file
• Parameters accumulate: Step N = original + param1 + param2 + ... + paramN
• Intermediate JPGs for validation only, never participate in next step processing

Directory Structure

workspace/
└── image-editor-work/
    ├── {filename1}/        # Independent working directory for image 1
    │   ├── temp/
    │   │   ├── step01.jpg
    │   │   ├── step02.jpg
    │   │   └── params/
    │   │       ├── step01_params.txt
    │   │       └── step02_params.txt
    │   └── final/
    │       └── {filename1}_final.jpg
    ├── {filename2}/        # Independent working directory for image 2
    │   ├── temp/
    │   │   ├── step01.jpg
    │   │   └── params/
    │   │       └── step01_params.txt
    │   └── final/
    │       └── {filename2}_final.jpg
    └── ...

Naming Conventions:

• Working directory name: {original filename (without extension)}
• Intermediate files: step{NN}.jpg
• Parameter files: step{NN}_params.txt
• Final files: {original filename}_final.jpg

Notes

• Processing RAW format requires ImageMagick with RAW decoding support
• High-resolution images take longer to process
• Recommend processing preview images first to confirm effects

Editing Step Quality Standards (Agent Guidelines)

Multimodal Check Points

When reviewing intermediate JPGs, watch for:

Step Type	Check Points	Standard Met
Contrast Adjustment	Highlight/shadow details, overall clarity	Shadow details present, highlights not blown, image has depth
Color Toning	Color temperature, saturation, color cast	Matches target style (e.g., teal-orange has cyan-blue shadows + warm yellow highlights)
Vignette	Corner darkening degree, transition smoothness	Corners noticeably darker, edge-to-center transition smooth, not jarring
Sharpening	Edge clarity, noise control	Subject edges clear, no obvious aliasing or noise increase
Grain	Grain uniformity, natural feel	Film texture natural, doesn't destroy details, grain evenly distributed
Destructive Editing	Color banding, noise, color patches from excessive parameters	Natural color transitions, no banding, no noise beyond added grain, no artifacts

Parameter Adjustment Strategy

• Insufficient effect: Increase parameter intensity (e.g., contrast from 15 to 25)
• Excessive effect: Decrease parameter intensity (e.g., vignette from 80% to 60%)
• Color cast: Adjust hue or change tint fill color
• Local issues: Consider if regional processing is needed

Important Reminders

1. Quality first: All intermediate JPGs must use -quality 100
2. Preserve original: Strictly prohibited from overwriting or modifying original images
3. Parameter recording: Write to parameter file immediately after step达标 to prevent loss
4. Always start from original: Each step starts from original file, stacking all previously verified params
5. Intermediate JPGs for validation only: Never use intermediate JPGs as input for next step
6. Final generation: Apply all parameters from original in one pass to generate final image