Prompt Helper Suite v1.1

A standalone semantic prompt workflow for AI image generation and animation

Prompt Helper Suite is a standalone, local-first prompt-building tool for creating, organizing, saving, reusing, and optimizing structured prompts for modern AI image generation and animation systems.

It is not a ComfyUI extension, not a ComfyUI node pack, not a renderer, and not locked to any single generation program.

At its core, Prompt Helper Suite is a copy-and-paste prompt workflow.

You create and organize the prompt inside Prompt Helper Suite, then copy it into the tool you want to use, such as ComfyUI, GPT image workflows, Nano Banana, Grok, Gemini, Flux, Z-Image, Z-Image Turbo, text-to-video tools, or other natural-language image and animation systems.

Optional API support can send prompts directly to ComfyUI, but this remains a convenience feature. Prompt Helper Suite continues to work as a completely standalone prompt system.

What's new in v1.1

Version 1.1 is a substantial update focused on workflow cleanup, local LLM integration, database safety, backup protection, and easier scene creation.

New in v1.1

• New Scene Director

• Generic Local LLM Optimizer through Ollama

• Support for selecting installed Ollama models

• Automatic LLM VRAM unloading

• New Luna the Watchdog database protection system

• Automatic local database snapshots

• Database integrity and unexpected shrink detection

• Recovery safeguards for damaged or unexpectedly reset databases

• New Luna Backup Assistant

• Guided external backup-folder setup

• Improved Backup / Restore workflow

• Safer SQLite database backups

• Improved navigation and workflow consistency

• Cleaner portable distribution

The goal of v1.1 is not to change how Prompt Helper fundamentally works, but to make the existing workflow safer, cleaner, easier to navigate, and more reliable.

Why it was created

Prompt Helper Suite was created to solve a practical problem: consistent prompting becomes difficult when the same character, outfit, pose, scene, lighting, tattoo, jewelry, anatomy, material, and camera details have to be rewritten manually every time.

Small details drift easily.

Eye color changes.

Hairstyles shift.

Body proportions become inconsistent.

Tattoos disappear or move.

Outfits lose construction details.

Lighting conflicts with the scene.

The same character slowly stops being the same character.

Prompt Helper Suite reduces that repetition by turning prompt elements into reusable semantic presets.

Create a character once.

Create an outfit once.

Create a pose once.

Create a scene once.

Create a tattoo setup, lighting setup, camera framing, animal, or recurring entity once.

After saving, these elements can be selected again with a few clicks and combined into new scenes without rebuilding the entire prompt from scratch.

What it does

Prompt Helper Suite acts as a structured semantic workflow layer above external renderers and AI tools.

It prepares clean, reusable, readable, structured prompts that can be copied into the program of your choice.

The suite can be used for:

• character consistency

• reusable outfit systems

• scene building

• fashion prompting

• cinematic image generation

• visual storytelling

• text-to-image workflows

• text-to-animation workflows

• recurring characters and environments

• structured material and anatomy descriptions

• multi-character prompting

• local LLM-assisted prompt optimization

• renderer-agnostic copy-and-paste workflows

Click-and-play workflow

Prompt Helper Suite is built around guided selection instead of constant manual rewriting.

Users can select saved characters, outfits, poses, scenes, lighting, camera setups, tattoos, jewelry, anatomy details, animals, companion entities, and material presets from the interface.

The suite then assembles the selected elements into a structured prompt.

Within a small number of clicks, a complete storytelling scene can be built with:

• one or more characters

• independent poses

• clothing and outfit presets

• anatomy details

• tattoos

• skin details

• jewelry

• animals or companion entities

• environment

• lighting

• camera framing

• material behavior

The final prompt remains visible and editable.

Nothing is hidden behind the interface.

You can inspect the generated prompt, change it manually, optimize it with a local LLM if desired, and copy the result into your preferred generation system.

Structured prompt blocks

Prompt Helper Suite builds prompts through semantic blocks such as:

[SCENE]

[LOOK / STYLE]

[SUBJECT]

[SKIN / TAN]

[HAIR STYLE]

[OUTFIT]

[POSE]

[TATTOO]

[SKIN DETAIL]

[JEWELRY]

[ANATOMY]

[LIGHTING]

[CAMERA & FRAMING]

This keeps prompts readable, modular, reusable, and easier to maintain.

Instead of one long unstructured prompt, the output remains organized as a clear scene description.

Reusable preset database

The suite uses a local SQLite preset database.

Users can save and reuse:

• characters

• outfits

• poses

• environments

• tattoos

• hairstyles

• jewelry

• anatomy presets

• camera setups

• lighting setups

• animals and companion entities

• garment systems

• material systems

• cinematic collections

Once something is saved, it becomes available throughout the workflow without needing to be rewritten manually.

Your own preset collection can therefore grow together with your workflow.

Independent entity and pose handling

Each character or entity in a scene can keep its own independent description and pose behavior.

One character can stand.

Another can sit.

Another can swim.

A dog can stand nearby.

A cat can sit on the floor.

Each entity can keep its own:

• pose

• posture

• movement

• outfit

• anatomy

• behavior

• scene role

This helps reduce prompt confusion in multi-character scenes and allows more complex scenes to remain readable.

Scene Director

Version 1.1 introduces the Scene Director.

Scene Director provides a separate workspace for developing scene ideas with a local Ollama LLM.

Instead of manually constructing every scene description from scratch, the user can provide the scene concept and let the selected local LLM help expand it into a more complete scene description.

The generated result remains under user control and can be copied or brought back into the normal Prompt Helper workflow.

Scene Director uses the same local Ollama setup as the LLM Optimizer, so no separate cloud service is required.

Local LLM Optimizer

Prompt Helper Suite includes optional local LLM optimization through Ollama.

The optimizer is no longer tied specifically to Gemma.

Users can select an available Ollama model and use it as a controlled semantic prompt editor.

The optimizer can help:

• reduce repetition

• preserve semantic block structure

• preserve character identity

• preserve triggerwords

• preserve anatomy

• preserve material relationships

• resolve simple conflicts

• reduce token weight

• improve readability

The original prompt remains available, allowing the user to compare the original and optimized versions before deciding what to use.

Optimizer behavior is controlled by editable rules, and a default ruleset is included.

Because Ollama runs locally, prompt optimization can remain completely local.

Automatic LLM VRAM unloading

Local LLMs can occupy a significant amount of GPU VRAM.

This can become inconvenient when Prompt Helper, Ollama, ComfyUI, and an image-generation model share the same GPU.

Version 1.1 therefore includes:

Auto unload LLM after each task

When enabled, Prompt Helper asks Ollama to release the LLM after an optimization or Scene Director task has finished.

This frees GPU memory for applications such as ComfyUI and Z-Image.

The option can be disabled in Settings when the user prefers to keep the LLM loaded for faster consecutive requests.

Auto unload is enabled by default.

Luna the Watchdog

Version 1.1 introduces Luna the Watchdog, a database protection layer built into Prompt Helper Suite.

Luna was created to protect the part of Prompt Helper that often becomes the most valuable over time: the user's own preset database.

Luna can monitor the active database for problems such as:

• database corruption

• unexpected database shrinkage

• unexpectedly disappearing presets

• sudden category loss

• accidental replacement by a small or newly initialized database

Luna performs database health checks and keeps information about the last known healthy state.

Automatic SQLite snapshots provide additional recovery points.

If Luna detects a database state that looks seriously wrong, the protection system is designed to avoid blindly continuing with potentially destructive database operations.

The protection follows the currently active user-data location and does not depend on a fixed drive or installation path.

This keeps Prompt Helper portable while still protecting existing user data.

Luna Backup Assistant

Luna is not only a watchdog.

She can also help users set up a proper backup location.

After the first few successful starts, if no external backup location has been configured, Luna can offer assistance.

Users remain completely in control.

I have a folder

Choose an existing backup folder yourself.

Help me choose

Luna checks the available storage drives and suggests a suitable location.

When multiple drives are available, Luna can recommend keeping backups on a different drive from the active Prompt Helper database.

The recommendation is only a recommendation.

The user can:

• select another drive

• use an external drive

• use a USB stick

• change the suggested folder

• change the folder name

• configure everything manually instead

Luna never forces a backup location and never creates the suggested folder without user confirmation.

The selected backup location is shared with the normal Backup / Restore system, so there is only one backup configuration to manage.

Safer Backup and Restore

Version 1.1 improves database backup behavior.

Prompt Helper uses SQLite-aware database snapshotting rather than relying only on copying an active database file.

This is particularly important for databases operating with SQLite WAL journaling.

Backup archives can contain the application data required for restoration while avoiding unnecessary temporary runtime database files.

The goal is simple:

Your presets should be easy to create, easy to move, and much harder to lose.

Layered material and anatomy logic

One of the main design goals of Prompt Helper Suite is to describe visual details as layered physical relationships rather than loose keywords.

The suite treats skin, tattoos, anatomy, clothing, transparency, moisture, jewelry, and layered garments as connected visual layers.

For example, a tattoo remains part of the skin even beneath pantyhose, mesh, lace, soaked cotton, or wet translucent clothing.

The tattoo does not become a print on the fabric.

Instead, the visibility is described through the material layer above it.

Thin or wet fabric may soften, diffuse, or partially reveal the tattoo while keeping it attached to the body underneath.

The same logic can be used for:

• tattoos beneath pantyhose

• tattoos beneath wet clothing

• skin detail beneath translucent fabric

• jewelry pressed against wet fabric

• layered garments over anatomy

• fabric compression

• wet skin

• moisture and transparency behavior

The user does not need to manually rebuild these layer relationships every time.

The structured preset system helps maintain them consistently.

Image and animation workflows

Prompt Helper Suite was originally built around image prompting, but structured scene descriptions can also be useful for text-to-animation and text-to-video workflows.

Consistent descriptions of character identity, clothing, anatomy, tattoos, material behavior, environment, lighting, and camera framing can help support visual continuity across generated motion.

The generated prompts can be used in external systems such as:

• ComfyUI

• GPT image workflows

• Nano Banana

• Grok

• Gemini

• Flux workflows

• Z-Image workflows

• Z-Image Turbo workflows

• text-to-video tools

• other natural-language renderers

Prompt Helper does not require any of these systems specifically.

Use whichever renderer or generation workflow fits your needs.

Import, export, and modules

Presets can be exported and shared with other users.

When importing preset packs, the suite can check for existing presets and avoid unnecessarily duplicating existing entries.

The system also supports data-driven module packs.

Modules and presets are based on readable data rather than executable plugins.

Shared module data can be inspected in a normal text editor before import.

No hidden executable plugins.

No runtime code injection.

No black-box module behavior.

Local-first design

Prompt Helper Suite is designed as a local-first workflow.

Presets, databases, modules, optimizer settings, prompt libraries, Scene Director input, and local LLM workflows remain under the user's control.

No cloud dependency is required for the core application.

Ollama integration is optional.

ComfyUI integration is optional.

Prompt Helper Suite itself remains a standalone semantic prompt-building system.

Use it with the tools you already have.

About version 1.1

Version 1.1 is primarily a stability, workflow, safety, and local-LLM update.

The focus has been on improving the existing foundation before adding larger new systems.

Prompt Helper Suite will continue to evolve, but the central idea remains unchanged:

Create reusable visual ideas once, combine them freely, keep the resulting prompt readable, and use it with the AI generation tools you prefer.