angel291592 / caddiff

git diff for CAD assemblies — two STEP files in, a picture of what changed plus a machine-readable change list out.

alexsprojects0 / Pen-plotter

A 180mm × 106.5mm pen plotter driven by four 28BYJ-48 stepper motors. It can run on any microcontroller with at least 12 output pins, but all my code is written for the Arduino Uno.

alanhuangyoo / OpenJev

OpenJev: open-source, local alternative to Jev. System-One decision models: typed questions in, calibrated probabilities out; general decisions and browser-agent steps.

dashs20 / CRAP

CRAP: Cycloidal Reducer Algorithm & Parametrizer. Parametric cycloidal gearbox generator (build123d): give it bearings, motor, ratio and eccentricity; get collision-checked, toleranced STEP/STL parts and a live 3D web GUI.

wavtechyukky / LeapSinger

Fast singing-voice acoustic model: harmonic + noise excitation refined into a mel by a 1-step rectified flow (bundled NHVSing vocoder). Multi-speaker + style control; ONNX/OpenUTAU export.

Saganaki22 / ComfyUI-Hyperflow

ComfyUI node pack for HyperFlow [Video Rebirth's] 8-step LoRA for MiniMax-H3 (video + audio), ported onto ComfyUI's native MiniMax-H3 model. No ComfyUI core files are modified; no custom sampler is needed.

wangyuxun6699 / learn_claude_code_langchain

Rebuild Claude Code's agent harness step by step with LangChain and LangGraph — from agent loops and tools to permissions, subagents, context compaction, memory, and recovery.

manifoldor / jumper-replica

Independent Jumper geometry recovery and assembly references: STL, faceted STEP, Blender exploded views, and pilot-fit 3MF. Physical replica not yet validated.

Chuf-H / jev-tree

Jev-native probability tree and graph runtime for verifiable multi-step decision making.

sepiablue-ai / ComfyUI-MiniMax-H3-TaoMate-C3

TaoMate-H3 3-step acceleration package for ComfyUI on an RTX 4070 12GB: 4-bit (W4A4) FFN weights, 10% sparse attention and extra INT8 weight caching, with Windows setup and sample workflows.

hikarioyama / ParallelHue

Truthful scheduler-step and streaming visualization for local LLM inference

shootthesound / ComfyUI-KMS

A KSampler that shows you N seeds before committing: cheap probe pass, clickable thumbnails on the node, then finishes only the seed you pick - exact continuation, no wasted steps. Model agnostic.

biyuhe3442-cmd / ComfyUI-NB-H3-HyperStep

—

beyondworks / castra

Execution posture harness for Claude Code. Your coding agent stops one step early; Castra is the step.

Boki-Lee / MD-Master

Unified MD reference library: a 0–6 step workflow with runnable CNT ion-transport and umbrella-sampling free-energy templates.

waltstephen / UMM-Reflection

Learning Native Reflection in Unified Models: reflection SFT and multi-round Flow-GRPO RL for BAGEL

SharkSpicy-NLP / Beyond-Factual-Knowledge

[EMNLP 2026] Beyond Factual Knowledge: Benchmarking and Learning Step-Level Procedural Rule Reasoning in Large Language Models

trustabl / agent-opentelemetry-rules

Agent OpenTelemetry (otel) rules pack: enriches opentelemetry with agent telemetry, steps, access rights and policy to explain agent failure.

stephrobert / terraform-dsoxlab-training

Lab pédagogique public de la formation Terraform (parcours du blog).

filonux / Mi-Lanzador

Graphical macro builder for Linux Mint. Chain steps: open apps, files, sites, commands; place windows; click; press keys; type text. Save them as one-click Desktop/menu shortcuts. Like AutoHotkey, on top of xdotool/wmctrl. One self-installing Bash script, ES/EN interface, GPLv3.

step-out / Motion-Omni

—

MTAI-HMU / NAC_audio_tools

Encode, decode and plot audio with 10 neural audio codecs and audio VAEs (DAC, EnCodec, SNAC, Stable Audio Open, Music2Latent, ACE-Step and more), all through one interface. Useful for looking inside each model's latent space and comparing them side by side.

Blakeem / ComfyUI-ContextAnchoredTileRefine

Seamless tiled refinement and upscaling for ComfyUI. The VL nodes step all tiles on one shared latent, conditioned by vision tokens and per-tile captions instead of a prompt, so objects never repeat across tiles. Refine whole images or just a masked region. Any GUIDER model on the base node; Krea 2 on the VL nodes. Tested at 8K+.

kadirnar / drifting-tts

One-step Turkish text-to-speech trained with a drifting objective (Kyutai learned-temperature recipe)

neu-vi / ViRDM

ViRDM: Taming Representation Distribution Matching for Few-Step Causal Video Generation