A durable async control plane for Codex Desktop that handles long-running tasks and Plan Mode workflows with retry safety and local orchestration support.
An AI-powered skill for Codex that transforms course materials into an adaptive Obsidian learning system with a personalized, feedback-driven workflow and comprehensive exam review cycle.
An open-source SDR plugin that fully demodulates Japan's satellite-based J-ALERT emergency broadcast system, recovering original alert data directly from RF signals offline.
An autonomous multi-agent AI research system that replicates a PhD workflow with real ML experiments, literature grounding, and anti-reward-hacking safeguards, all on a single-GPU budget.
An open-source, museum-catalog-quality digital collection of Chinese traditional patterns with structured data, accessible online and downloadable for free.
An open-source Windows utility enabling massive FPS boosts and high-quality upscaling through technologies like LSFG frame generation, AMD FSR, and Nvidia NIS.
An open-source collection of 52 meticulously crafted AI video ad prompts spanning 10 formats, designed for performance marketers with multi-beat timing and fidelity guards.
This repository provides a highly optimized, narrow-phase collision detection library for convex primitives, achieving approximately 100x speedup over a reference implementation through LLM-assisted code generation and rigorous testing.
CodexPro enables ChatGPT Developer Mode to function as a local coding agent by connecting it to a workspace through MCP tools for inspection, editing, and safe verification.
A toolchain for designing Figma title cards and generating procedural animated patterns in-browser, exportable as layered PNG sequences for Blender compositing.
A Python crypto wallet analyzer bypassing DeBank fraud detection through proxy rotation while supporting multi-threaded address and seed phrase checks across major blockchains.
An open-source unified image generation and editing model family that achieves competitive performance with smaller training data through systematic improvements in understanding, data quality, and training pipelines.