Search

A modular Retrieval-Augmented Generation (RAG) pipeline for Python.

通用、模块化、可扩展的 TypeScript RAG SDK

检索模块，支持查询变换、多种搜索策略与后处理

Qdrant 向量数据库适配器

Google 嵌入模型适配器

Voyage AI 嵌入模型适配器

嵌入模型提供者抽象接口，内置 OpenAI 适配器

索引管道，编排文档加载、清洗、切块、嵌入与存储

生成模块，提供提示词模板与多种生成策略

评测模块，支持检索、生成与端到端评测

知识图谱模块，提供实体抽取、图谱构建与图谱增强检索

RAG SDK 核心模块，提供全局类型定义、Pipeline 编排器与路由

Google Gemini LLM 适配器

HelixID middleware for LangChain and LangGraph — verifiable credential enforcement at agent execution boundaries with per-tool scope checking.

Fast, local, zero-setup RAG with no embedding model — BM25 + the LLM you plug in.

Storage backend adapters for storagesdk. Import the adapter you need via a subpath, e.g. `@storagesdk/adapters/fs`.

A thin, declarative React wrapper around OpenSeadragon: hooks, components, and a small plugin mechanism.

Deterministic PDF/DOCX parser for RAG / LLMs — Rust core with Python, Node & WASM bindings. Placeholder, under construction.

OpenCode plugin that discovers LiteLLM models at runtime.

Developer-first fairness regression testing for LLM applications.