Projects Recently Starred by Experts

Starred 5 hours ago by

Donny Kang(Author of SWC; Maintainer of Turbopack). Also starred by

Patrick von Platen(Author of Hugging Face Diffusers; Research Engineer at Mistral),

Junyang Lin(Core Maintainer at Alibaba Qwen),

Boris Cherny(Creator of Claude Code; MTS at Anthropic), and

22 more.

claude-code by anthropics

5.7%

85k

Agentic coding assistant for your terminal

Understands and navigates your codebase via natural language commands
Executes routine tasks, explains code, and manages Git workflows
Integrates with IDEs and GitHub for seamless interaction
Enhances coding speed and efficiency through AI assistance

Created 1 year ago

Updated 17 hours ago

Starred 12 hours ago by

Chaoyu Yang(Founder of Bento). Also starred by

Wing Lian(Founder of Axolotl AI) and

Anton Bukov(Cofounder of 1inch Network).

turboquant_plus by TheTom

6956.9%

LLM KV cache compression for efficient local inference

Achieves 4.6x KV cache compression using PolarQuant and Walsh-Hadamard rotation.
Delivers zero speed penalty on Apple Silicon, matching q8_0 performance.
Integrated into llama.cpp with Metal GPU acceleration for seamless deployment.
Maintains high quality and retrieval accuracy (NIAH) at extreme compression levels.

Created 1 week ago

Updated 18 hours ago

Starred 14 hours ago by

Shizhe Diao(Author of LMFlow; Research Scientist at NVIDIA). Also starred by

Hiroshi Shibata(Core Contributor to Ruby),

Dan Guido(Cofounder of Trail of Bits), and

Elvis Saravia(Founder of DAIR.AI).

codex-plugin-cc by openai

N/A

IDE integration for AI code review and task delegation

Conduct standard or adversarial code reviews directly within your development workflow.
Delegate complex tasks like bug investigation and code fixing to an AI assistant.
Leverages local Codex CLI for seamless integration and utilizes existing configurations.
Enables steerable reviews to pressure-test design choices, tradeoffs, and assumptions.

Created 2 days ago

Updated 22 hours ago

Starred 17 hours ago by

Didier Lopes(Founder of OpenBB). Also starred by

Peer Richelsen(Cofounder of Cal.com),

Gregor Zunic(Cofounder of Browser Use),

Bryan Helmig(Cofounder of Zapier), and

7 more.

sim by simstudioai

0.3%

27k

Open-source AI agent workflow builder

Build/deploy LLMs connecting to tools via intuitive interface.
Self-hosted via Docker, supports local models (Ollama).
Uses Next.js, PostgreSQL/Drizzle ORM, Better Auth, ReactFlow.

Created 1 year ago

Updated 15 hours ago

Starred 20 hours ago by

Simon Suo(Cofounder of LlamaIndex). Also starred by

Jerry Liu(Cofounder of LlamaIndex).

liteparse by run-llama

60.8%

Fast, local document parsing and screenshotting for AI

Extracts text with bounding boxes from PDFs, Office docs, and images locally.
Integrated Tesseract.js OCR and supports external OCR servers for robust text recognition.
Leverages PDF.js for parsing and converts various file types to PDF for unified processing.
Generates high-quality page screenshots essential for visual information extraction by LLM agents.

Created 1 month ago

Updated 1 day ago

Starred 21 hours ago by

Jay V(Cofounder of SST). Also starred by

George Hotz(Author of tinygrad; Founder of the tiny corp, comma.ai).

opencode-claude-auth by griffinmartin

21.9%

581

Seamless Anthropic API authentication for OpenCode

Uses existing Claude Code OAuth credentials, no separate login or API key needed.
Manages token caching, background refresh via CLI, and cross-platform credential syncing.
Integrates with macOS Keychain and local JSON files for secure, reliable access.
Automates token refresh and handles API request lifecycle for uninterrupted use.

Created 1 week ago

Updated 22 hours ago

Starred 21 hours ago by

Chaoyu Yang(Founder of Bento). Also starred by

Jeremy Howard(Cofounder of fast.ai) and

Zhuohan Li(Coauthor of vLLM).

marlin by IST-DASLab

0.5%

FP16xINT4 kernel for fast LLM inference

Achieves near-ideal 4x speedups up to medium batch sizes (16-32 tokens).
Optimizes GPU resource utilization: memory, cache, tensor cores.
Includes GPTQ algorithm version for Marlin-compatible Llama2 compression.
Provides a torch-module for easy integration with existing models.

Created 2 years ago

Updated 1 year ago

Starred 1 day ago by

Max Howell(Author of Homebrew). Also starred by

David Cournapeau(Author of scikit-learn).

sentrysearch by ssrajadh

97.0%

Semantic search and retrieval over video footage

Embeds video chunks using Gemini or local models for vector-based semantic search.
Retrieves precise, trimmed video clips matching natural language queries.
Supports local Qwen3-VL backend for private, on-device processing.
Optional Tesla metadata overlay and still-frame skipping for efficiency.

Created 2 weeks ago

Updated 2 days ago

Starred 1 day ago by

Didier Lopes(Founder of OpenBB).

oh-my-codex by Yeachan-Heo

121.3%

Multi-agent orchestration for OpenAI Codex CLI

Enhances Codex CLI with role-based agents, workflow skills, and team orchestration.
Manages complex tasks via staged execution, persistent state, and memory.
Extends functionality with a plugin-based hook system for additive automation.
Provides structured decomposition, lifecycle management, and operational controls for Codex.

Created 1 month ago

Updated 11 hours ago

Starred 1 day ago by

Jeff Hammerbacher(Cofounder of Cloudera).

agent-sop by strands-agents

1.5%

861

AI agent workflows defined by natural language SOPs

Guide AI agents through complex, multi-step tasks with markdown instructions.
Enable consistent and reliable execution using parameterized inputs and constraints.
Integrates with MCP Server and Anthropic Skills for broad AI system compatibility.
SOPs can be authored by AI agents, supporting AI-driven workflow creation.

Created 4 months ago

Updated 3 days ago

...