展示 HN：灵根 – 为你的 AI（光标、Zed、Claude）设计的本地优先型记忆层

展示 HN：灵根 – 为你的 AI（光标、Zed、Claude）设计的本地优先型记忆层
Show HN: Linggen – A local-first memory layer for your AI (Cursor, Zed, Claude)

原始链接: https://github.com/linggen/linggen

## Linggen：代码的 AI 记忆 Linggen 是一款免费的、本地优先的应用，旨在为你的 AI（如 Cursor、Zed 或 Claude）提供代码库和项目知识的记忆。它解决了 AI “盲目” 的问题，通过索引你的代码并将架构决策存储为可搜索的 Markdown 文件来实现。主要功能包括：**持久记忆**，允许 AI 回忆过去的决策；**跨项目智能**，实现跨多个项目的学习；以及 **系统地图**，可视化依赖关系以进行更安全的重构。重要的是，所有数据处理——使用 LanceDB 进行索引和向量搜索——都*本地*在你的机器上进行，确保隐私并无需账户。 Linggen 通过 MCP 与兼容的 IDE 集成，允许你直接查询它（“调用 Linggen MCP，了解…”）。它采用开源模式，个人用户免费，并为需要支持和未来企业功能的团队提供商业许可，例如 SSO 和团队同步。

## Linggen：本地AI记忆层 Linggen是一款新的、本地优先的工具，旨在为Cursor、Zed和Claude等AI工具提供持久的上下文。它由一位对反复向LLM解释复杂系统感到沮丧的开发者创建，Linggen *本地*索引项目文档和代码——无需账户——实现即时回忆，并消除“冷启动”问题。 Linggen使用Rust和LanceDB构建，提供“团队记忆”功能以共享上下文，并提供可视化地图以了解文件依赖关系。它通过VS Code扩展激活，快速加载用户工作日所需的的相关信息。开发者强调，Linggen的优势在于它提供可搜索、易于获取的上下文*在*AI工作流程中，从而提高效率并减少冗余解释。该项目是开源的，可在GitHub上找到：[https://github.com/linggen/linggen](https://github.com/linggen/linggen)。

原文

The free and local app for your AI’s memory.

Linggen indexes your codebases and tribal knowledge so your AI (Cursor, Zed, Claude, etc.) can actually understand your architecture, cross-project dependencies, and long-term decisions.

Website • VS Code Extension • Documentation

Traditional AI chat is "blind" to anything you haven't manually copy-pasted. Linggen bridges this "context gap" by providing:

🧠 Persistent Memory: Store architectural decisions in .linggen/memory as Markdown. AI recalls them via semantic search.
🌐 Cross-Project Intelligence: Work on Project A while your AI learns design patterns or auth logic from Project B.
📊 System Map (Graph): Visualize file dependencies and "blast radius" before you refactor.
🔒 Local-First & Private: All indexing and vector search (via LanceDB) happens on your machine. Your code and embeddings never leave your side. No accounts required.

Install the CLI in seconds and start indexing:

curl -sSL https://linggen.dev/install-cli.sh | bash
linggen start
linggen index .

Windows & Linux support coming soon.

💬 How to use it with your AI

Once Linggen is running and your project is indexed, simply talk to your MCP-enabled IDE (like Cursor or Zed):

"Call Linggen MCP, find out how project-sender sends out messages, and ingest it."

"Call Linggen MCP, load memory from Project-B, learn its code style and design pattern."

"Load memory from Linggen, find out what is the goal of this piece of code."

📂 The Linggen Ecosystem

linggen: The core engine and CLI runtime.
linggen-vscode: VS Code extension for Graph View and automatic MCP setup.
linggensite: (This Repo) The landing page and documentation site.
linggen-releases: Pre-built binaries and distribution scripts.

Linggen is open-source under the MIT License.

100% Free for Individuals: Use it for all your personal and open-source projects.
Local-First: Your code and your "memory" never leave your machine.
Commercial Support: If you are a team (5+ users) or a company using Linggen in a professional environment, we ask that you support the project's development by purchasing a Commercial License.

For more details on future enterprise features (SSO, Team Sync, RBAC), visit our Pricing Page or get in touch via email.