Search

Biology grant reasoning skill for China's funding context (NSFC, MOST) — mechanism-centered diagnosis, reviewer-aware evaluation, multi-level project adaptation

Structured scientific reasoning skill for AI agents — Claude Code, OpenClaw, Hermes, Pi-Mo, Codex, SkillsMP

FlashHead: vLLM Plugin for Fast Language Model Head Inference

Autonomous engineering framework. A brain agent reads a decision-consequence dashboard, thinks, and delegates to sub-agents in git worktrees. Self-improving — the system writes its own scheduling rules, triage logic, and evaluation cadence from its own decision history. Pluggable: works with any LLM backend (Claude, Codex).

Production-grade Android best practices for AI coding assistants. Works with Cursor, Claude Code, GitHub Copilot, OpenAI Codex, and Windsurf.

Client and Tools for LLMs

Fast Python linter for catching LLM-generated anti-patterns

Local voice dictation: Whisper STT + Claude LLM post-processing

PromptLine SDK — governed LLM execution, local or remote

LangChain integration for CoordiNode — GraphStore backed by graph + vector + full-text

Fast BPE tokenizer for LLMs

High Quality and Blazing Fast ArXiv paper search — 928K papers, SPECTER2 or Gemini-2 embeddings

Python SDK for the bigRAG platform

Structured image analysis using LLMs. Define a Pydantic model, send an image, get structured data.

LLM-friendly repo summarizer — indexes codebases into markdown summaries with incremental git-diff updates

Install local middleware proxies into Claude Code

Mezzo Guard is a Python SDK designed to add a protective layer to LLM applications attacks such as jailbreaking and prompt injection, powered by custom-trained models adapted to recent trends. Mezzo Guard can be easily added into your pipeline, offering great protection with little effort.

None

Turn anything you read into structured notes. Schema-driven extraction from URLs, PDFs, EPUBs, images, and clipboard using any LLM.

Adversarial security testing framework for LLM-powered applications