Stars
One llama.cpp flag unlocks +33-39% decode speed for Qwen3.8-27B on consumer GPUs. The MTP head already ships inside your GGUF. Recipe, paired benchmarks, probe tool.
29 editorial diagram types for Claude Code. Self-contained HTML + SVG. No shadows, no Mermaid-slop.
KISS Sorcar: Keep it Simple, Stupid AI General Purpose Agent Framework for Long Horizon Tasks and AI Discovery
Distilled Code Reviewer skills/soul from 32k~ email from the kernel newsletter
Codex-native architect orchestration with Luna and Terra implementation lanes and mandatory fresh Sol review.
Local-first healthcare AI: clinical NER & HIPAA PII de-identification that runs 100% on-device. 2,200+ medical models, 21 languages, Apple MLX + Python, no cloud, no patient data leaving your netwo…
Open-source Conductor.build alternative. Runs the real claude/gemini/codex CLIs in real terminals, no SDK middleman
role-model is a protocol for assigning the right model for the right job. Use local and cloud AI together, or route between several cloud providers.
28 eval-informed mental models and critical-thinking skills for Claude Code, GitHub Copilot, Codex, Cursor, and other Agent Skills-compatible tools
A skill to stop your coding agent from burying the answer. ADHD-friendly output.
The meta-skill that builds and improves all your skills, including itself. Watches your work sessions, captures corrections and judgement calls, and turns them into skill improvements automatically…
A personal financial advisor that runs locally. Plaid CLI + SQLite + Claude Code skills. Fiduciary-grade advice grounded in your real data.
Conformance and capability test suite for LLM inference engines — surface coverage, spec conformance, model capability, and an MTP/performance benchmark.
Claude Cowork skill/plugin to help with Australian Tax Returns
AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI
DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm
Give text-only LLMs vision. A tiny OpenAI-compatible proxy that lets reasoning models (DeepSeek, Qwen, GLM…) see images by querying a separate vision model through tools: look, OCR, scan, crop, com…
AI on consumer Blackwell — a field guide for the RTX 50-series (sm_120). Fixes, footguns, and honest measurement from one RTX 5090.
autonomous red teaming platform; multi-agent offensive-security meta-harness
A TTS that fits in your CPU (and pocket)
Reduce GPT-5.5 token usage 75.5% by supervising a 27B local model
Lonkero - Wraps around your attack surface. Professional-grade scanner for real penetration testing. Fast. Modular. Rust.
poteto's Claude Code agent-style skill — concise responses, deliberate subagents, unslopped prose, verified work (MIT, by Lauren Tan)