Seth's AI Tools: A Unity based front-end for ComfyUI and Ollama (and others) that does things like generate photos and movies, Twine games, quizzes, posters and more.

License: BSD style attribution, see LICENSE.md

Download the latest: V2.51 (Jan 30th, 2026) AI Tools Client (Windows, 62 MB)

Need an old version? The last pre-V2 version (that still supports Auto1111) can be downloaded here.

NOTE: This software is hard to use because it kind of assumes you already know a lot of about ComfyUI and how to use local LLMs like llama.cpp.

If this is all weird and strange, this is probably not the place for you to start. Although, you could just enter an OpenAI API key for both image creation and LLM I guess and this app could do some fun stuff, but the main point of this is to do everything in your home lab locally.

Features

• It's not a web app, it's a native locally run Windows .exe (Unity with C#)
• It's kind of like lego, it can run ComfyUI workflows that chain together. You need to understand ComfyUI and workflow tweaking as you'll still need to install all the needed custom nodes and models there. (you do have ComfyUI Manager installed, right?!)
• Drag and drop images in as well as paste images from the windows clipboard
• Built-in image processing like cropping and resizing, mask painting
• Pan/zoom with thousands of images on the screen
• Built to utilize many ComfyUI servers at once
• Privacy respected - does not phone home or collect any statistics, purely local usage. (it does check a single file on github.com to check for newer versions, but that's it)
• Includes "experiments", little built-in games and apps designed to test AI. May be broken
• AI Guide feature harnesses the power of AI to create motivational posters, illustrated stories or whatever
• Adventure mode allows playing a realtime illustrated adventure (kind of like AI Dungeon) or generate quizes and choose your own adventures that can be exported to self contained playable HTML versions.

Recent changes

V2.51 (Jan 30th, 2026)

• Adventure mode has basic markup display support, quote coloring (can be disabled in options)
• Input tweaked, the ENTER key just stopped working app-wide on my version, this fixed it
• Unity version updated
• LLM repetition value can be controlled

V2.50 (Jan 13th, 2026)

• Added vision LLM support - use @llm_add_image to send images to vision-capable LLMs
• New script commands: @upload, @parse_llm_prompts, @stopjob for more flexible preset workflows
• Added drag multi select, can delete or move things as a group
• Job script got a lot of upgrades, it can do things like look at an image via a VLLM, decide the best way to edit it, generate a prompt for that via LLM, and do the edit to end up with a kind of "we have nano banana at home" experience. Also supports generating multiple text prompts for a single thing, useful for certain ComfyUI workflows that have multiple stages that need them
• Fixed bugs with Adventure mode's "Compact" button, also its "Stop after X" option
• The whole autopic system is more powerful and can use llms and comyui workflows with its jobscript in a standalone way, previously it had to be pared with a Preset. Individual GPU/servers conected can override the AutoPic to do their own thing if needed now
• Various stability, tweaks, and bug fixes

V2.12 (Dec 18th, 2025)

• A Christmas miracle, I actually updated this! I figured enough changed that people might want to try it, although I'm still kind of not planning to seriously support this as it's just too niche and hard to make easy to use
• We can now only work with normal ComfyUI workflows, hallelujah! No more "API" version exporting!
• Adventure mode now has story and interactive story "AutoPic" modes, it's a newer technique that works better, instead of generating image data with the story, it separately automatically generates image descriptions based on the story. (controlled with AutoPic.txt)
• Erased a lot of presets and added some new ones, the main ones I use these days are Z-Image and Wan
• CrazyCam is replaced by "SpookyCam", a photobooth thing I made, check the preset for it for more info. Normal crazycam is too broken to use right now
• LLM support overhauled, it now has a fully GUI based config and improved support for llama.cpp and Ollama
• While local AI is the main target, ChatGPT 5.2, GPT Image 1.5, Claude 4.5, etc. are also supported (requires API keys to use)

Known issues

• CrazyCam disabled
• A lot of the mask/editing stuff I haven't used in a while, not sure if it still works
• Things tend to get broken if I haven't used that particular feature in a while. It's probably the app, not you

Screenshots

Media (mostly outdated videos of the app)

Setup

Your ComfyUI server needs to be started using the --listen parm, so the API can be accessed.

After running it, a config.txt will be made. There is a config edit button inside the app, it's kind of self-documented, just go through it and update as necessary.

Setting up with ComfyUI

First, install ComfyUI and get it rendering stuff.

Then install my Workflow to API Converter Endpoint via ComfyUI Manager

This node allows us to only work with normal workflow json files, and never have to export "API" versions. The API conversions will happen under the hood automatically when needed, makes life much better. Click its page above for help.

It will work if you don't, but this allows modified or new workflows to work as it can convert them to "API" versions on the fly when needed.

Next, just for a test to make sure the workflows included with AITools are going to work, inside ComfyUI's web GUI, drag in aitools_client/ComfyUI/FullWorkflowVersions/text_to_img_flux.json or any others. The neat thing about ComfyUI is it will read this and convert it to its visual workflow format, ready to run. (you might want to change the prompt from <AITOOLS_PROMPT> to something else during testing here) - Click Queue. Does it work? Oh, if you see an Image Loader set to the file "<AITOOLS_INPUT_1>" you'll need to change that to a file on your ComfyUI server if you want to test.

You'll probably see a bunch of red nodes and errors - no problem! Make sure you have ComfyUI-Manager installed, you can use it to install any missing nodes. You'll probably have to track down various model files though, but at least when you try to render it will shows exactly the filenames that are missing. (look for red boxes around certain nodes)

Adjust it until it works (change paths or models or whatever you need, you could even start with a totally different workflow you found somewhere else), and make sure the prompt is set to <AITOOLS_PROMPT>. (<AITOOLS_NEGATIVE_PROMPT> can be used if your workflow has a place for that too) Then do Workflow->Export if you wanted to save your own. (note: we no longer use "API" versions at all, thank god, just normal workflows and dynamically convert them as needed)

So you don't have to create custom workflows for every checkpoint/filesize etc, you can use AITools' "@replace" to change any part of a workflow before it's sent to ComfyUI's API, but after it's been converted to an "API" version of the workflow. You'll see it used in various presets.

Check discussions for some more info here

List of keywords that can be used in ComfyUI workflows so AITools can modify things

Keyword	Description
<AITOOLS_PROMPT>	Main prompt used to generate whatever (image or movie). Same as <AITOOLS_PROMPT_1>
<AITOOLS_PROMPT_1> - <AITOOLS_PROMPT_8>	Extended prompts for multi-segment generation (e.g., different prompts for each 5 seconds of a 20 second movie). AITOOLS_PROMPT_1 falls back to AITOOLS_PROMPT if empty
<AITOOLS_NEGATIVE_PROMPT>	Some image generators use this, but a lot ignore it
<AITOOLS_AUDIO_PROMPT>	Audio prompt for audio things
<AITOOLS_AUDIO_NEGATIVE_PROMPT>	What you DON'T want to hear (e.g., to exclude music from audio)
<AITOOLS_SEGMENTATION_PROMPT>	For SAM3 segmentation - like "head" and the head gets selected
<AITOOLS_INPUT_1> - <AITOOLS_INPUT_4>	For images. Use Load Image (Path) (a V.H.S. node) anywhere a normal image is needed. Supports up to 4 image inputs via @upload command

List of Job Script commands

Job scripts are used in presets to chain workflows and manipulate variables. Each line can contain a workflow name followed by commands separated by @. It's kind of rough, but you can call LLMs with text, get something, run workflows, tweak the image and use it on other workflows, etc.

Basic Syntax

workflow_name.json @command|parm1|parm2| @command2|parm1|

Example: img_to_img.json @copy|prompt|segmentation_prompt| @resize_if_larger|x|1024|y|1024|aspect_correct|1|

Commands

Command	Parameters	Description
@copy	source|dest|	Copy text/image between variables
@add	source|dest|	Append text to a variable
@resize_if_larger	x|width|y|height|aspect_correct|0 or 1|	Resize image only if larger than specified
@resize	x|width|y|height|aspect_correct|0 or 1|	Force resize to specified dimensions
@upload	source|inputN|	Upload image to ComfyUI input slot (source: image/temp1-temp3, dest: input1-4
@fill_mask_if_blank	(none)	Fills the alpha mask if empty
@no_undo	(none)	Disables undo for this operation
@stopjob	(none)	Stop adding more jobs after this script completes
@llm_prompt_reset	(none)	Reset LLM conversation history
@llm_prompt_set_base_prompt	text| or multi-line	Set base system prompt for LLM
@llm_prompt_add_from_user	text| or multi-line	Add user message to LLM conversation
@llm_prompt_add_from_assistant	text| or multi-line	Add assistant message to LLM conversation
@llm_prompt_pop_first	(none)	Remove first interaction from history
@llm_prompt_add_to_last_interaction	text| or multi-line	Append text to last message
@llm_add_image	source|	Add image to next LLM message for vision models (source: image/temp1-temp3)
@parse_llm_prompts	(none)	Parse SET_PROMPT1 through SET_PROMPT8 from llm_reply into prompt variables
@Comment	text|	Human-readable comment (ignored by parser)

Multi-line Block Syntax

The LLM prompt commands above support multi-line text using block syntax. Instead of putting text on the same line with pipes, put the command alone and the content on following lines, ending with @end:

command @llm_prompt_add_from_user
Format your response as:
SET_PROMPT1: [first 5 seconds description]
SET_PROMPT2: [seconds 5-10 description]
SET_PROMPT3: [seconds 10-15 description]
SET_PROMPT4: [seconds 15-20 description]
@end

Single-line syntax still works: command @llm_prompt_add_from_user|short text|

Variable Names (for @copy/@add)

Variable	Description
prompt	Current job's main prompt
global_prompt	Global prompt field in UI
prepend_prompt	Text prepended to prompts
append_prompt	Text appended to prompts
negative_prompt	Negative prompt for generation
audio_prompt	Audio generation prompt
audio_negative_prompt	Audio negative prompt
segmentation_prompt	Prompt for SAM segmentation
llm_prompt	Prompt sent to LLM
llm_reply	Response received from LLM
requirements	Requirements string
image, image1	Current image (synonyms - for copying to/from temp slots)
temp1, temp2, temp3	Temporary image storage slots
temp_text1 - temp_text4	General-purpose text storage buffers for job scripts

Special Line Prefixes

Prefix	Description
command	Execute commands without running a workflow (just runs the @ commands on that line)
call_llm	Call the LLM with the current prompt, response stored in llm_reply

Note: Workflow filenames like img_to_img_*.json or video_to_video_*.json are just naming conventions. To upload images for img2img workflows, use the @upload command explicitly.

AutoPic Presets

AutoPic presets (e.g., AutoPic.txt, AutoPicLTX20s.txt) are special preset job scripts used by Adventure mode to automatically generate image prompts from story text using an LLM. They're stored in the Presets/ folder and can be selected in the settings. Files starting with AutoPic are automatically detected and show in a drop down in the General Settings dialog. If @stopjob is used in one, it won't automatically try to run the main job script after it's done. (useful so the AutoPic can handle running the comfy workflow it specifically needs)

Disabling Commands

Add - at the start of a line to comment it out completely Add - at the beginning of a command to skip the next parameter

-workflow.json (comments out this whole line)
workflow.json -@disabled_command|parm| @enabled_command|parm|

Building from source

You only need to download the zip and run the .exe to use this, However, the source might be useful to generate a build for other platforms, fork or steal pieces to use for yourself. Go ahead!

• Requires Unity 6000.3.1f+
• Open the scene "Main" and click play to run
• Assets/GUI/GOTHIC.TFF and Assets/GUI/times.ttf are not included and might break the build because I was having trouble and switched some settings around that might require them now (dynamic vs static TMPro font settings...)

---

Credits and links

• Audio: "Chee Zee Jungle"

Kevin MacLeod (incompetech.com)

Licensed under Creative Commons: By Attribution 3.0

http://creativecommons.org/licenses/by/3.0/

• NotoSansCJKjp-VF font licensed under the Open Font License (OFL)
• Audio: JOHN_MICHEL_CELLO-BACH_AVE_MARIA.mp3 performed by John Michel. Licensed under Creative Commons: By Attribution 3.0

http://creativecommons.org/licenses/by/3.0/

• Written by Seth A. Robinson ([email protected]) twitter: @rtsoft - Codedojo, Seth's blog
• Special thanks to the ComfyUI and llama.cpp projects