Repository Index

A unified framework for building, evaluating, and recursively improving embodied agents. 🚀 Code release coming soon.

Reconquer the canvas: beautiful Tikz figures without clunky Tikz code

• tikz
• tikz-figures
• latex
• scientific-visualization

Make Gemini code like Claude Models. Forces it to inspect files before editing, plan changes, test its work, and back up every “done” with actual evidence.

• agent-skills
• antigravity-skills
• gemini
• prompt-engineering

Code and data for GeneGPT.

Code and Pretrained Models for Interspeech 2023 Paper "Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong Audio Event Taggers"

• audio
• audio-classification
• audio-processing
• audio-tagging

A model generator for energy system modelling and optimisation (LP/MILP).

• energy
• energy-system
• modelling
• modelling-framework

Iterative Linear Quadratic Regulator with auto-differentiatiable dynamics models

• trajectory-optimization
• differential-dynamic-programming
• theano
• dynamics-models

Build AI text detectors whose decisions are mechanistically grounded and human-auditable (EMNLP 2026)

• ai-text-detection
• explainable-ai
• interpretability
• mechanistic-interpretability

GEARS is a geometric deep learning model that predicts outcomes of novel multi-gene perturbations

TranslatedOne zero-dependency MCP gateway CLI for every server and agent skill. Token optimization, tool discovery and context compression: 71,929 -> 581 tokens (-99.2%, measured), schemas stay out of context. One config for Claude Code, Codex, Cursor, every agent. 561KB, pure Python. | Zero-dependency CLI: manages all MCP tools and skills, and tool discovery saves 99.2% of tokens.

• mcp
• claude
• claude-code
• cursor

Real-time generative rendering for interactive worlds, from any visual reference.

• game
• renderer
• video-generation
• game-engine

The first open-source model for code-driven video creation.

Learn a codebase by re-deriving its design decisions yourself. A Claude Code skill.

• agent-skills
• claude-code
• claude-code-plugin
• claude-skills

Naive-N0.5-Flash is an open-weight 309B MoE model with 15.5B active parameters, built for coding and AI R&D.

Text mode diagrams using UTF-8 characters and fancy colors

Provides utilities for Postgres database schema versioning. This is being superseded by work done at https://codeberg.org/perseas/perseid.

• python
• postgres
• schema-evolution

News reader with tags, scoring, and LLM

• feed
• feed-aggregator
• feed-reader
• news

Wirehair : O(N) Fountain Code for Large Data

Upload once to create multiple, relative PRs. Productivity-focused git tools with patch revision tracking, rebase detection and more. Uses python and git plumbing commands.

• code-review
• git
• github
• version-control

Reproduction of SuperMimic (arXiv:2610.28480): learned differentiable world model + BPTT policy for G1 humanoid motion tracking, on mjlab/MuJoCo Warp

Hyperparameter optimization and feature selection for scikit-learn using evolutionary algorithms. A modern alternative to GridSearchCV and RandomizedSearchCV.

• python
• scikit-learn
• machine-learning
• artificial-intelligence

TranslatedOpen-source mobile vertical short-form micro drama app — short drama distribution platform and drama CMS, with full source code. Open-source vertical-screen micro short drama distribution platform: short drama system / short drama distribution / short drama source code fully open, aimed at taking short dramas overseas. Content providers join and upload, the platform reviews and distributes, with episode unlocking and settlement.

• micro-drama
• ott
• short-drama
• svod

Lightweight Claude Code statusLine: 5h/7d rate-limit usage, reset countdowns, model + context window, prompt-cache age — one line, 3 styles × 9 themes, daemon fast-mode

• ai-tools
• claude-ai
• developer-tools
• monitoring

Pathology Language and Image Pre-Training (PLIP) is the first vision and language foundation model for Pathology AI (Nature Medicine). PLIP is a large-scale pre-trained model that can be used to extract visual and language features from pathology images and text description. The model is a fine-tuned version of the original CLIP model.

• artificial-intelligence
• clip
• pathology
• vision-and-language