AI Tools

gpt 4ogemini 2 0 groqopenaigemini

Train GPT-2-class language models from scratch in C/CUDA

GPT-2-class language model built from scratch in C/CUDA - hand-written forward/backward passes, BPE tokenizer, FlashAttention, no ML libraries.

Connect LLMs to your shell with a secure Bash CLI wrapper

Bash4LLM+ is a single, auditable Bash script CLI wrapper for Groq's OpenAI-compatible Chat Completions API.

Route prompts to local or cloud models by complexity

Wayfinder scores prompt structure offline, with no model call, to route cheap prompts local and hard ones to cloud.

openaianthropicgemini

91 8 days ago

universal claudecursoraider

Keep your Mac awake only while AI coding agents are working

Adrafinil is a macOS menu bar app that keeps your Mac awake through a closed lid only while an AI agent works.

90 12 days ago

Compress files using overfitted neural networks and arithmetic coding

pym-particles is an experimental neural compressor that overfits a small transformer on one file plus arithmetic coding to compress it byte-by-byte.

91 last month

Embed a themeable AI chat widget into any website

Persona is a themeable, framework-free TypeScript AI chat widget you drop into any website, with streaming, voice, tool calls, and artifacts.

openaivercellangchain

claudegemini 2 0 claudegithubgit

Edit markdown wikis with AI-powered WYSIWYG editor

OpenKnowledge is a WYSIWYG markdown editor for knowledge bases and LLM wikis, with Claude and Codex editing built in.

86 5 days ago

opentelemetrylangfusearize

Optimize AI agents using production trace analysis

HALO is an RLM-based agent optimizer that turns production traces into concrete fixes for your agent harness.

Monitor Claude Code sessions with live spend tracking and approval

Pulse is a local dashboard for Claude Code showing spend and session state, with phone approval for tool calls.

claude claudentfy

91 15 days ago

claudegpt 4o huggingfaceopenaianthropic

Run LLM inference with multimodal support and API serving

mistral.rs is a fast Rust LLM inference engine with zero-config model loading, multimodal support, and built-in agentic tools.

Generate parametric 3D CAD models from text descriptions

CADAM turns natural-language and image prompts into parametric CAD models, exportable as STL, SCAD, or DXF, in your browser.

openscadsupabasereact

91 11 days ago