Search

PetaLLM allows single 4GB GPU card to run 70B large language models without quantization, distillation or pruning. 8GB vmem to run 405B Llama3.1.

Claude Skill for finding professors and research opportunities

Claude code ansible skill

skillmd project.

Local-first code graph builder with hybrid vector + graph search

rrag - retrieval-reasoning augmented generation engine

CLI harness for CLIProxyAPI - proxy server with OpenAI/Gemini/Claude/Codex compatible APIs

This proj_flow extension aids with builds covered with github.com/mbits-libs/coveralls

Python wrapper for claudectx — reduce Claude Code token usage by up to 80%

CLI tool for the Web Resurrect API

Official SDD methodology skills for HarnSpec

An adaptive learning and coprocessor architecture for [Claude Code](https://docs.anthropic.com/en/docs/build-with-claude/claude-code), built as an extension to [GSD (Get Shit Done)](https://github.com/gsd-build).

A Starlight plugin that turns your documentation into AI-installable skills

CLI tool to scaffold new projects from boop profiles and capabilities

Explore and map any CLI tool deeply. Parse --help trees, mine shell history for repeated workflows, surface hidden flags via LLM archaeology, and suggest new agent skills from your usage patterns.

A Pi package for persistent markdown wikis with source capture, generated metadata, linting, and an LLM wiki-maintainer skill.

Agent SDK for SkillAuditor — submit, verify, and check skills onchain in one call

OpenCode plugin that learns from skill usage and suggests relevant skills automatically

Multi-account OAuth rotation for OpenAI Codex with auto model discovery

Native web search for Pi using OpenAI's Responses API via Codex backend (same as Codex CLI)