🍯 Masterphub Trending

If you've ever wanted to know what the rest of the world is actually prompting for, Masterphub Trending is the node for it. It pulls the most-used prompts in any category from Masterphub's 160,000+ prompt library, returns the single most popular one ready to generate from, and also hands you a ranked list with usage counts. It's one of four nodes in the small masterphub-comfyui pack - Search, Load by ID, Random, and this one - a new, low-traffic pack whose main public presence is the author's own announcement post on r/comfyui.

How it works

Like every node in the pack, it's a thin HTTPS call to the Supabase backend behind masterphub.io, using an anon key baked into the source. The query orders prompts by unlocked_count descending - "unlocked" is Masterphub's counter for how often a prompt has been used - and caps the result at your limit. That's the whole mechanism. No models, no inference, no API key, Python standard library only.

Inputs and outputs

Two inputs:

• category - all, image, video, writing, marketing, business, coding, agents, audio, 3d. Pick image if you're generating images; category all mixes in writing, business, and coding prompts that are useless in a diffusion encoder.
• limit - 1–10 (default 5), how many entries to rank.

Two outputs, both STRING:

• top_prompt - the single most-used prompt's full text. This is what you wire into CLIP Text Encode.
• trending_list - a numbered list of the top results, formatted like 1. [12345] Title (9876 uses). The bracketed ID feeds Masterphub Load by ID if you want to pull any of them deterministically.

Why you'd reach for it

Trending is the pack's answer to "what actually works?" - popularity as a proxy for quality, which is crude but not crazy. The uses count in the list is genuinely useful signal: a prompt with 9,000 unlocks is a community-validated starter; one with 12 is a shrug. And unlike Masterphub Random, Trending is deterministic - same category, same limit, same result - which makes it the shareable one. You can drop it into a workflow, and it produces a stable, reproducible prompt instead of a moving target.

The honest caveats: ranking is by raw popularity, so you'll get broad crowd-pleasers rather than anything subtle, and prompt quality from any public catalog is a mixed bag - some are tag-style, some are prose, and they won't behave identically across model families. It's also a network call on every run, so offline you'll see an Error: ... string on the outputs, and an empty category returns the string No results. rather than crashing.

Install

ComfyUI Manager → Install Custom Nodes → search Masterphub → Install → restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/stavjadam-collab/masterphub-comfyui

then restart and look for the 🍯 nodes under the Masterphub menu. Watch out: the README's own clone command points at github.com/masterphub/masterphub-comfyui, a different org than the real repo (stavjadam-collab/...). If that line fails for you, use the URL above or let ComfyUI Manager do the fetching. No requirements.txt, no model downloads - one of the friendlier installs in the ecosystem.