compress-images
by rameerezcompress-images automates image optimization for web and SEO performance. It moves source JPG, PNG, or GIF files into an originals/ folder, converts them to WebP, iterates until files are under 100KB, and updates content references for a safer batch workflow.
This skill scores 78/100 and is worth listing: it gives agents a clear, actionable image-compression workflow with a defined trigger, default path, and repeatable output target. For directory users, that means it should be easy to install and use for real web/SEO image optimization tasks, though it is not a fully packaged toolset and still leaves some adoption details implicit.
- Explicit trigger and default use case: compress images for web/SEO performance, with a default target directory when no path is provided.
- Operational workflow is concrete: preserve originals in an originals/ folder, convert JPG/PNG/GIF to WebP, and iterate until files are under 100KB.
- Strong agent leverage from command-level guidance and code examples, reducing guesswork versus a generic prompt.
- No install command or helper files are provided, so users must rely on the SKILL.md instructions rather than a broader package/setup workflow.
- Reference updating is mentioned, but the repository evidence shows limited constraints/practical guidance, so edge cases and content-file formats may require manual judgment.
Overview of compress-images skill
What compress-images does
The compress-images skill automates image optimization for web use: it takes a directory of JPG, PNG, or GIF files, preserves the originals, and outputs WebP versions tuned for smaller file sizes and better page performance. If you need a compress-images skill that turns a rough image folder into SEO-friendly assets without hand-tuning every file, this is built for that job.
Best fit for this workflow
Use compress-images when your goal is to reduce image weight for content pages, landing pages, or blog posts where load speed matters. It is most useful when you already know the target folder and want a repeatable compression pass instead of a generic “optimize images” prompt.
What makes it different
The key distinction is the repo’s workflow discipline: it keeps originals in an originals/ subfolder, writes WebP output beside them, and iterates until files are under a size target. That makes the compress-images guide practical for production content workflows, not just one-off conversion.
How to Use compress-images skill
Install and invoke it
Use the compress-images install flow with the repo’s CLI pattern:
npx skills add rameerez/claude-code-startup-skills --skill compress-images
Then invoke it with a directory argument when possible. The skill is designed to work on $ARGUMENTS, and if you do not provide one it defaults to app/assets/images/content/.
Give it the right input
The compress-images usage pattern works best when you specify a folder that contains only the images you want processed and where updated references can be safely changed. Strong input looks like: compress-images ./content/posts/post-42/images. Weak input is just “compress my images,” because the skill needs a concrete path and a clear scope boundary.
Read the repo in the right order
Start with SKILL.md because it contains the actual compression process and size target. Then inspect the surrounding skill metadata and any linked repo docs if present. For this repo, the main decision points are the process rules: preserve originals, convert to .webp, compress iteratively, and update references in content files.
Workflow tips that affect output
Use the skill when you want a safe batch workflow: originals moved aside first, then compressed assets generated, then references updated after verification. To get better results, tell the agent whether image dimensions should stay intact, whether small files still need conversion, and whether the destination folder is content-managed or code-managed. Those details affect how aggressively the skill should compress and how carefully it should rewrite references.
compress-images skill FAQ
Is compress-images only for WebP conversion?
No. WebP conversion is the main output, but the real purpose of the compress-images skill is to reduce image weight for faster delivery while preserving source files. If your stack still needs PNG or JPG elsewhere, this may not be the best fit.
When should I not use this skill?
Do not use it if you need pixel-perfect archival preservation in the working directory, if the folder contains mixed media that should not be bulk processed, or if you are not ready to update content references after conversion. The skill assumes a deliberate image-optimization pass, not casual editing.
Is it beginner-friendly?
Yes, if you can identify a target directory and understand that originals will be moved into originals/. It is more structured than a plain prompt, which makes it easier to use safely, but you still need to know which files should be included.
How is this different from a generic prompt?
A generic prompt may compress images, but compress-images gives a defined workflow: target folder, preserved originals, iterative compression, and reference updates. That makes the result more predictable for Image Editing and site-performance tasks.
How to Improve compress-images skill
Specify the success threshold
The biggest quality lever is the target you want the agent to optimize for. This skill defaults to under 100KB per image, but if your page layout, image count, or CMS constraints differ, say so up front. A stronger prompt names the folder, the file types, and whether the 100KB target is strict or approximate.
Provide context about layout and usage
compress-images works better when you tell it where the images appear and how much visual fidelity matters. For hero images, product shots, and inline screenshots, the right tradeoff differs. If you only say “optimize these images,” the agent has to guess whether to preserve detail or chase smaller files.
Watch for common failure modes
The main risks are over-compression, wrong-scope processing, and broken references after extension changes. To reduce those risks, ask the agent to report before/after sizes, list files moved into originals/, and confirm which references were updated. That makes the compress-images guide easier to verify in one pass.
Iterate with a clearer second pass
If the first output is too large or too soft, improve the prompt by naming the specific files that missed the target and the acceptable tradeoff. For example: “Re-run only the three hero images; keep text legible; prioritize size over quality only if the file is still above 100KB.” That gives the compress-images skill enough direction to refine output without recompressing everything blindly.