需先自行安装并运行 Ollama 及拉取模型;嵌入模型需另行准备;安装与部署步骤在 docs/setup.md 中,README 未直接给出命令
local-rag
jonfairbanksGPL-3.0活跃维护在本地网络内用开源大模型完成文件摄取与检索增强生成,数据不出内网、不依赖第三方服务。
Ingest files for retrieval augmented generation (RAG) with open-source Large Language Models (LLMs), all without 3rd parties or sensitive data leaving your network.
README 未给出安装命令,详见 docs/setup.mdREADME 未说明具体显存或 GPU 型号要求;模型由本地 Ollama 或 Hugging Face 提供,硬件门槛取决于所选模型。
本地离线 RAG 工具,用 Ollama 模型索引本地文件、GitHub 仓库和网站,数据不出内网
在本地网络内用开源大模型完成文件摄取与检索增强生成,数据不出内网、不依赖第三方服务。
核心能力
- 接入本地 Ollama 聊天模型进行问答
- 支持 Ollama 或本地 Hugging Face 嵌入模型
- 可摄取本地文件、GitHub 仓库和网站三类来源
- 通过 LlamaIndex 流式输出 RAG 回答
- 设置保存在浏览器本地,支持导出聊天历史
容易踩的坑
- v2.0.0最新正式版2026-05-18 · 2026-05-18 06:23 UTC
What's Changed
- Update Dockerfile by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/76
- Bump version to v2.x by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/75
Full Changelog: https://github.com/jonfairbanks/local-rag/compare/v1.4.4...v2.0.0
- v1.4.3Bug Fixes & Dependency Updates2024-07-05 · 2024-07-05 01:11 UTC
查看更新说明
What's Changed
- Switch to cache_resource for Document Index by @JoepdeJong in https://github.com/jonfairbanks/local-rag/pull/54
- Fixes application crash when models not found by @cybertechy in https://github.com/jonfairbanks/local-rag/pull/58
- Bug Fixes and Dependency Updates by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/62
New Contributors
- @JoepdeJong made their first contribution in https://github.com/jonfairbanks/local-rag/pull/54
- @cybertechy made their first contribution in https://github.com/jonfairbanks/local-rag/pull/58
Full Changelog: https://github.com/jonfairbanks/local-rag/compare/v1.4.2...v1.4.3
- v1.4.1Security Update2024-05-04 · 2024-05-04 06:02 UTC
查看更新说明
Updates
This update resolves a critical command injection vulnerability present in
llama-index, an underlying dependency.https://nvd.nist.gov/vuln/detail/CVE-2024-3271
What's Changed
- Security Update by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/45
Full Changelog: https://github.com/jonfairbanks/local-rag/compare/v1.4.0...v1.4.1
- v1.4.0Bug Fixes2024-03-22 · 2024-03-22 21:34 UTC
查看更新说明
What's Changed
- Contributing section by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/27
- update pipfile by @bananya-ml in https://github.com/jonfairbanks/local-rag/pull/29
- Windows Bug Fix by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/30
- Update Pipfile.lock by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/32
- Update Pipfile by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/38
- Migrate to Llama-Index settings by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/41
New Contributors
- @bananya-ml made their first contribution in https://github.com/jonfairbanks/local-rag/pull/29
Full Changelog: https://github.com/jonfairbanks/local-rag/compare/v1.3.3...v1.4.0
- v1.3.3Cuda Support2024-03-07 · 2024-03-07 03:10 UTC
查看更新说明
What's Changed
- Allow embeddings to use cuda devices by @jonfairbanks in https://github.com/jonfairbanks/local-rag/pull/26
Full Changelog: https://github.com/jonfairbanks/local-rag/compare/v1.3.2...v1.3.3
用户点评
登录用户的真实使用记录,可按版本和硬件筛选还没有人点评。第一个说出真实体验的人最有价值。
Local RAG

Offline, open-source retrieval augmented generation.
Ingest local files, GitHub repositories, and websites for retrieval augmented generation (RAG) with local Ollama models. Local RAG keeps chat, embeddings, and indexed source content on your machine or network.
Features:
- Local Ollama chat models
- Ollama or local Hugging Face embedding models
- Multiple ingestion sources:
- Local files
- GitHub repositories
- Websites
- Streaming RAG responses through LlamaIndex
- Browser-local settings persistence
- Chat history export
- Upload, URL, repository, and ingestion guardrails
Learn More: