# SilkStack — organize, search and view your AI images locally, with optional on-device AI. Free core, open source, v2.4 out now. — retained record

> Retained reddit record from reddit:comfyui.

Canonical: https://highsignal.app/data/records/6719257391106268

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:comfyui)  published 2026-10-09 07:12 ingested 2026-10-10 02:37

[original source ↗](https://www.reddit.com/r/comfyui/comments/1x1eqs3/silkstack_organize_search_and_view_your_ai_images/)

## retained text



```
Hey everyone,

My last post about this app got buried in the H3 posts and never really reached anyone, so I'm giving it another shot — hopefully it finds a wider audience this time.

SilkStack Image Browser is a desktop app for organizing your generated images and videos, and for viewing them the right way. It started as a fork of Image MetaHub, but I've been evolving it in my own direction for a while now, and I think it has its own distinct flavor at this point.

**What it does**

* Indexes your output folders (and watches them live while you generate) and reads the metadata inside the files — ComfyUI, A1111 and friends, including WebP.
* Search and filter by prompt, model, LoRAs, steps, CFG, sampler, seed, and more.
* A proper viewer: zoom with a minimap, Ctrl+F to search inside a prompt, and a compact mode that shows just the image so you can keep an eye on generations while working in another app.
* Similarity stacks group near-identical generations into one card and highlight the words that differ between prompts, so you can see exactly what changed.
* Drag an image straight into ComfyUI to reload its workflow.
* Videos and GIFs too (MP4/WEBM/GIF).

**The part I'm most excited about:** I've been using this app as my AI playground to implement local AI. Auto-tagging and semantic search run entirely on your own GPU — a little LLM writes the tags, and an embedding model lets you search by meaning instead of exact words. Describe "girl with a puppy in a palace" and it finds the image even when the prompt words are completely different, or in another language. It's heartening to see these features work with little LLMs.

The AI features are the premium part, unlocked with a 7-day free trial that becomes a monthly subscription (cancel anytime) — or a one-time lifetime license if you'd rather own it. AI loading is completely in your control: nothing loads until you turn it on from the interface, and you can eject the models anytime to hand your VRAM back.

Everything other than the AI features is fully open source — free for your agent's pleasure to play with. 🙂

One honest note: the app is primarily built around photos. Video support is there, and if you run into any bugs with videos, let me know — I can help fix them.

A lot of you liked my initial app, and I'm really curious to know what you think about this app now.

**Download (Windows / macOS / Linux):** [skkut.github.io/silkstack](https://skkut.github.io/silkstack/)

**Source:** [github.com/skkut/SilkStack-Image-Browser](https://github.com/skkut/SilkStack-Image-Browser)
```



document fetched 2026-10-10 00:41 · canonical https://reddit.com/r/comfyui/comments/1x1eqs3/silkstack_organize_search_and_view_your_ai_images
