Nox Prompt NSFW Designer

Nox Prompt NSFW Designer is the flagship node of the Nox Prompter pack, and it's the one people actually find. The pitch isn't "write me a dirty prompt" - it's "write me a dirty prompt with the guardrails already attached." You pick a pose, a wardrobe, a tone, and a heat level from curated dropdowns, and it hands you three strings: a prompt, a negative prompt, and safety notes you can keep as a compliance log. For an 18+ workflow on SDXL-lineage models, that's genuinely handy.

How it works

Everything in Nox Prompter runs on big curated vocabulary dictionaries in constants.py - no API calls, no LLM, no models to download. Designer resolves your dropdown selections into prompt fragments: pose_profile pulls from the pose dictionary, wardrobe_style from the wardrobe dictionary, and so on through tone_profile, lighting_setup, camera_framing, set_design, explicitness_level, and heat_profile. Each option maps to a prompt string and a summary, so the same selection that feeds the prompt also feeds a readable description.

The two toggles that matter are include_negative_baseline and include_safety_note, both on by default. The baseline appends a stock negative list (blurry, low detail, broken anatomy, awkward pose…), and the safety note appends the pack's default adult-safety text. The author's position is blunt: leave both on unless you've got verified policy tooling downstream. You can substitute your own copy via custom_safety_note and negative_prompt.

Every dropdown also has a Custom Input option that opens a matching custom_* field, so when the preset vocabulary isn't what you need, you can drop in your own phrasing instead of fighting the defaults. And like most nodes in the pack, preset_action / preset_name let you save, load, or list your configurations as JSON under presets/.

The inputs that actually matter

• pose_profile, wardrobe_style, tone_profile - the three required picks that do most of the work (16, 23, and 17 options respectively).
• explicitness_level and heat_profile - the "how far does this go" controls, from Soft Suggestive up to Fully Adult Nude.
• include_negative_baseline / include_safety_note - the guardrails. Leave them on.

Outputs: prompt (feed to your positive conditioning), negative_prompt, safety_notes, and preset_status (a human-readable string saying what the preset action did).

Installing it

Same story for the whole pack. In ComfyUI Manager, search "Nox Prompter" and install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/jamesk9526/NoxPrompter-ComfyUI-CustomNode

Restart ComfyUI and you'll have a NoxPrompter category on the right-click menu. There's no requirements.txt and no model download - it's pure stdlib Python, so install is painless.

Where people get burned

The README still points at an old ComfyNoxPrompter repo and folder name; the real repo is NoxPrompter-ComfyUI-CustomNode. The aggregator injects its own path on import, so a renamed folder mostly still works - but use the correct URL. Two notes on expectations: the pack's default negatives and prose style are built for SDXL-lineage models (Pony, Illustrious, NoobAI), where negatives still do real work - on LLM-encoded models like Wan 2.2 the negative box does far less. And the "safety" here is a guardrail in the text, not a content filter: this node doesn't block anything itself, it just makes sure the baseline terms and consent language ride along in every generation.