Image Resizer

Image Resizer & Converter

Why this skill exists

Image processing has many silent failure modes: wrong aspect ratio, quality loss, wrong color mode for format (RGBA can't save as JPEG), and EXIF data stripping. This skill handles all of them correctly.

When to use

• User uploads an image and wants it resized, cropped, converted, or compressed
• User wants thumbnails, WebP conversion, or batch processing of a folder

Step-by-step procedure

Step 1 — Inspect the image first

from PIL import Image
import os

path = "/mnt/user-data/uploads/photo.jpg"
img = Image.open(path)

print(f"Format: {img.format}")
print(f"Mode: {img.mode}")         # RGB, RGBA, L, P, etc.
print(f"Size: {img.size}")         # (width, height)
print(f"File size: {os.path.getsize(path) / 1024:.1f} KB")

Step 2 — Resize (preserve aspect ratio)

from PIL import Image

def resize_image(input_path, output_path, max_width=None, max_height=None, exact_size=None):
    img = Image.open(input_path)

    if exact_size:
        # Exact size — may distort
        img = img.resize(exact_size, Image.LANCZOS)
    elif max_width or max_height:
        # Preserve aspect ratio
        img.thumbnail(
            (max_width or 99999, max_height or 99999),
            Image.LANCZOS
        )

    # Fix mode before saving
    output_format = output_path.rsplit(".", 1)[-1].upper()
    if output_format == "JPG":
        output_format = "JPEG"
    if output_format == "JPEG" and img.mode in ("RGBA", "P"):
        img = img.convert("RGB")  # JPEG doesn't support transparency

    img.save(output_path, format=output_format, quality=85, optimize=True)
    print(f"Saved: {output_path} ({img.size[0]}×{img.size[1]})")

# Examples
resize_image("input.jpg", "output_800w.jpg", max_width=800)
resize_image("input.png", "thumbnail.png", max_width=200, max_height=200)
resize_image("input.jpg", "exact.jpg", exact_size=(1920, 1080))

Step 3 — Crop to square (thumbnail style)

def crop_to_square(input_path, output_path, size=512):
    img = Image.open(input_path)
    w, h = img.size
    min_dim = min(w, h)

    # Center crop
    left = (w - min_dim) // 2
    top = (h - min_dim) // 2
    img = img.crop((left, top, left + min_dim, top + min_dim))
    img = img.resize((size, size), Image.LANCZOS)

    img.save(output_path)
    print(f"Square crop saved: {output_path} ({size}×{size})")

Step 4 — Convert format

def convert_format(input_path, output_format="webp", quality=85):
    img = Image.open(input_path)
    base = input_path.rsplit(".", 1)[0]
    output_path = f"{base}.{output_format}"

    if output_format.lower() == "jpeg" and img.mode in ("RGBA", "P"):
        img = img.convert("RGB")

    img.save(output_path, format=output_format.upper(), quality=quality)
    
    in_size = os.path.getsize(input_path) / 1024
    out_size = os.path.getsize(output_path) / 1024
    print(f"Converted: {output_path} ({in_size:.0f}KB → {out_size:.0f}KB)")
    return output_path

Step 5 — Batch processing

import os
from pathlib import Path

def batch_resize(input_dir, output_dir, max_width=800):
    Path(output_dir).mkdir(parents=True, exist_ok=True)
    exts = {".jpg", ".jpeg", ".png", ".webp", ".gif", ".bmp"}

    files = [f for f in Path(input_dir).iterdir() if f.suffix.lower() in exts]
    print(f"Processing {len(files)} images...")

    for i, file in enumerate(files, 1):
        out = Path(output_dir) / file.name
        resize_image(str(file), str(out), max_width=max_width)
        print(f"  [{i}/{len(files)}] {file.name}")

Edge cases

| Situation | Fix | |-----------|-----| | RGBA → JPEG error | img.convert("RGB") before saving as JPEG | | Palette mode (P) | img.convert("RGB") first | | Animated GIF | Use ImageSequence.Iterator — resizing just takes frame 0 | | EXIF rotation | ImageOps.exif_transpose(img) before resizing | | Very large image (>50MP) | Use Image.open with draft() to load at reduced size | | Output dir missing | os.makedirs(output_dir, exist_ok=True) |

Output format

Always report:

• Original dimensions and file size
• Output dimensions and file size
• Compression ratio (if converting/compressing)
• Output file path

Move final files to /mnt/user-data/outputs/ so the user can download them.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: prasad-nimbalkar
• Source: prasad-nimbalkar/claude-agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.