轻量 AI 工作流平台Lightweight AI workflow platform· 2026
LOOM
轻量 AI 工作流与 RAG 知识库Lightweight AI workflow and RAG knowledge base
LOOM-PLUS 的前身之一，专注 SQLite 单机部署和低门槛工作流学习体验。A lightweight predecessor to LOOM-PLUS focused on single-machine deployment and low-friction workflow learning.
案例摘要Case summary
产品洞察Product insight
不是所有 AI 工具都需要一开始就上重型基础设施。对学习和验证场景来说，SQLite 单机部署能显著降低理解和启动门槛。Not every AI tool needs heavy infrastructure from day one. For learning and validation scenarios, SQLite single-machine deployment significantly lowers the bar for understanding and startup.
项目简介Overview
LOOM 是基于 NestJS + Vue 3 的轻量 AI 工作流平台，聚焦单机部署、工作流基础能力和知识库检索。它为后续 LOOM-PLUS 的融合版提供了轻量化方向和模块验证。LOOM is a lightweight AI workflow platform built on NestJS + Vue 3, focused on single-machine deployment, basic workflow capabilities, and knowledge-base retrieval. It provided the lightweight direction and module validation for the later merged LOOM-PLUS.
技术栈Tech stack
结果证据Evidence
本地仓库 remote 指向 qrx-joe/LOOM，最近 50 条提交均为 qrx-joe。The local repo remote points to qrx-joe/LOOM, with the latest 50 commits all authored by qrx-joe.
项目结构包含 frontend、backend 和启动脚本。The project structure includes frontend, backend, and startup scripts.
作为 LOOM-PLUS 的融合来源之一，验证了轻量部署方向。As one source for the LOOM-PLUS merge, it validated the lightweight deployment direction.
关键决策Key decisions
AI 工作流学习项目如果一开始依赖 PostgreSQL、Redis、Docker，全链路启动门槛会偏高。If an AI workflow learning project depends on PostgreSQL, Redis, and Docker from the start, the full-stack startup bar is too high.
用 SQLite 单机模式降低部署复杂度，先验证工作流和知识库核心体验。Use SQLite single-machine mode to lower deployment complexity and validate the core workflow and knowledge-base experience first.
想看完整的代码和实现细节？Want the full code and implementation details?
在 GitHub 上查看View on GitHub