AI applications

An open-source infinite-canvas workbench for AI film and video creation, combining multimodal generation, storyboard sequencing, asset management, and agent workflows for creative teams.

A desktop AI workspace that converts screenshots, handwriting, and PDFs into editable LaTeX formulas with live preview and Office integration.

macOS utility that divides the MacBook desktop into four tappable zones, locally classifying tap locations to trigger corresponding user-defined actions without sending data elsewhere.

An Iron Man-inspired holographic orb frontend for AI assistants, combining gesture recognition with 3D rendering for interactive voice and visual control.

AI college-application advisor for Chinese Gaokao candidates, combining public viewpoints with verified data, asking follow-up questions, tracing sources, and scoring universities and majors within a structured framework.

Serves scientists as a self-evolving AI research colleague with persistent memory and hundreds of skills, supporting long-term academic tasks while emphasizing verifiable, low-hallucination assistance.

Parses Counter-Strike demo recordings to extract highlights, funny fails, and meme-worthy deaths, then uses LLMs for witty scoring and commentary while automating OBS replay capture into finished clips.

AI short-video generator turns prompts into ready vertical videos for TikTok, Reels, and YouTube Shorts, providing scripting, voiceover, subtitles, templates, batch rendering, and multilingual output for creators.

Autonomous discovery platform connecting scientific concepts to executable analysis by extracting reusable principles and composing them into traceable, inspectable research ideas worth testing.

A local AI-driven roleplay kit linking iPhone camera feeds, an AI storyteller, and ESP32 feedback hardware.

Supports human-supervised brain-science research by coordinating multiple expert agents across literature review, data analysis, and report generation for end-to-end scientific workflows.

Automated subtitle tool for video creators that covers downloading, speech transcription, translation, and burned-in rendering in a single pipeline without manual intervention.

AI-assisted toolkit rebuilding flowcharts, architecture diagrams, and paper figures from images into editable Visio files, with SVG and PNG export for researchers and engineers.

A decoupled deep-research agent architecture that iteratively refines outlines through generation and critique loops, then renders web pages, posters, slides, and reports.

Command-line tool turning document collections into knowledge graphs by extracting entities and relationships with LLMs, supporting deduplication approval and interactive visualization.

A Python package and AI agent for querying arXiv research papers, enabling users to search literature, read sections, and obtain verifiable citations as if conversing directly with paper authors.

A minimal CLI and web UI for OpenAI GPT Image 2 generation, supporting text-to-image and image-to-image workflows, parallel generation, custom sizes, and dual API-key or ChatGPT OAuth authentication.

Supports AI-assisted video production from concept and planning through generation and editing, helping creators manage the full studio workflow toward finished publishable videos.

Local-first AI novel-writing workbench with built-in key management and multi-model workflows, supporting outlines, worldbuilding, and long-form drafting pipelines for fiction authors.

Dialogue-guided Chinese novel-writing engine supporting four-level outlines, style synthesis, foreshadowing DAG management, and long-text compression to help web-fiction authors maintain consistency.

Backend service for healthy-diet guidance built with Bun and TypeScript, supporting nutrition chat, food-image analysis, document ingestion, and grounded recommendations.

Collection of ComfyUI custom nodes providing low-cost third-party APIs for image and video generation, supporting calls to multiple large models with accompanying tutorials.

Free, open-source, fully local film-breakdown notebook that uses AI to extract story-lane timelines, structure trees, and emotion curves while users write notes alongside playback.

Clones voices, designs custom timbres, and builds multi-speaker dialogues through a Gradio web UI powered by Qwen3-TTS and VibeVoice, with Whisper-based automatic transcription support.

ComfyUI nodes and workflows turning Krea 2 into an instruction-based, identity-preserving image editor, using dual source-image injection for consistent people and scene edits.

Builds an AI-native product suite linking creation, collaboration, publishing, and operations in one platform with a visual workflow editor, community site, and admin console.

An open-source AI video studio that turns story ideas into finished videos with script, character, storyboard and rendering pipeline.

Local natural-language to shell command generator powered by a fine-tuned 1.5B Qwen2.5-Coder model. Running offline on CPU in about one second, it helps developers translate intent into reviewable commands for safer terminal use.

Offers a low-cost hybrid research assistant with a persistent kernel for scientists, combining inexpensive APIs to support scientific computation, analysis, and literature workflows.

Generates expressive speech from short clips to full-length audiobooks, cloning any voice from a 10-second reference while controlling emotion, pacing, and breathing for audiobook creators and voice designers.

Helps developers create API documentation through conversational vibe coding, automatically producing data models while offering built-in document viewing and mock API services for testing.

Local-first open-source AI video workbench combining external models and local ComfyUI, letting Claude Code, Codex, or Cursor direct storyboarding, generation, and timeline editing over MCP.

A local, offline AI audio suite combining speech transcription, synthesis, voice conversion, and music generation. Designed for creators and engineers needing professional-grade voice and music processing without cloud dependence.