Agent Skills Hub

reload

Clean up your current Claude Code session and automatically resume in a fresh, optimized terminal window.

paper-reproduce

Systematic methodology for reproducing published academic papers using provided data, including sample selection, statistical verification, and automated reporting.

differentiation-schemes

Generate finite-difference stencils, select optimal numerical schemes for PDEs/ODEs, and perform truncation error analysis to improve simulation accuracy.

goodqunbot

AI-powered WeChat message assistant. Retrieve recent messages from contacts/groups and send messages directly via AI chat. Windows 10/11 only.

claude-code

Standardized skill for Claude Code agents to dynamically query OpenRouter model recommendations and metadata via the Claudish CLI.

evaluation

Build systematic evaluation frameworks for AI agents using multi-dimensional rubrics, LLM-as-a-judge, and regression testing to measure performance, quality, and context engineering effectiveness.

mls

Unified local ML inference server for ASR, TTS, Translation, Image Generation, and Vision on Apple Silicon, powered by MLX.

math-extractor

Extracts mathematical content like definitions, theorems, and proofs from documents (PDF, MD, TEX, TXT) using AI-based cleaning and conversion.

llm-integration

A toolkit for building robust LLM integrations: API patterns, streaming, function calling, RAG pipelines, and cost-effective model routing.

massive-context-mcp

Process massive files and large codebases (10M+ tokens) by recursively chunking, sub-querying, and aggregating results to overcome LLM context limits.

debug-distributed

Debugging guide for AReaL distributed training issues, including hangs, NCCL errors, OOM, and numerical consistency in FSDP2/TP/CP/EP.

codeql

Perform deep security analysis on codebases using CodeQL for interprocedural data flow, taint tracking, and automated vulnerability detection across multiple languages.