Mnē-MCP
Mnē-MCP — 面向 opencode 的 RAG MCP 服务器
Mnē-MCP(源自希腊语 Mnēmosynē,记忆女神提坦;ē 中的长音符号 U+0113 标记了音译中的长元音)是一个为 opencode 实现完整 RAG(检索增强生成)管道的 MCP 服务器:它索引 Obsidian 仓库或文档文件夹,并使用用户自己的笔记作为来源回答助手的问题,同时引用每个使用片段的精确路径。
技术栈:#RAG #MCP #mcp-server #opencode #Obsidian #Qdrant #Ollama #DeepSeek #bge-m3 #Python
功能
增量索引 — 从磁盘扫描目录(
.md、.txt),通过mtime+md5检测变更,清理过时点(stale cleanup),并支持大容量卷(约 19 GB)。本地嵌入 — 通过本地 Ollama(
/api/embed)使用bge-m3模型,1024 维,多语言(含西班牙语),L2 归一化以正确适配余弦相似度。语义检索 — Qdrant(Docker),单一集合,通过 payload 过滤(
source、path、mtime、md5、heading_path),可配置相关性阈值。有据生成 — 仅当证据超过阈值时,才使用
deepseek-v4-flash(Ollama Cloud)生成回答;否则返回明确拒绝("no relevant documents found")。无证据,不生成。
Related MCP server: mcp-duckvault
架构
MCP tools (stdio) ──┐
├── RagService (core) ────▶ Qdrant (vectorial, Docker)
Admin page (:8310) ─┘ │
├── OllamaEmbedder (bge-m3 local)
└── LLMProvider (deepseek-v4-flash, Ollama Cloud)Python 3.14 · 官方 MCP SDK(
mcp==1.12.4)·qdrant-client·httpx7 个工具:
index·query·search·delete·list·stats·config本地管理页面(
http://127.0.0.1:8310):仪表盘、已索引文档、带进度的异步索引和搜索试验场。与 MCP 工具共享同一核心(零重复逻辑)。70 个离线测试套件(Qdrant
:memory:+ 模拟)+ 端到端冒烟测试。
开发者用法
docker compose up -d qdrant # Qdrant
ollama pull bge-m3 # embeddings locales
cp .env.example .env # OLLAMA_API_KEY + VAULT_ROOT
make install # instala el paquete (venv)
make admin # panel: http://127.0.0.1:8310在 opencode.json 中注册(mcp 部分,local 类型,命令 python3 mcp_rag/server.py 或入口点 mcp-rag-opencode)。
请求 | 行为 |
| 增量扫描,跳过未变更文件,批量上传向量 |
| 检索 → 重排序器/LLM → 带引用的回答( |
| 带 |
| Qdrant/Ollama 健康状态、LLM 状态、按来源计数 |
路线图
密集+稀疏混合搜索(BM25 + RRF)· cross-encoder 重排序 · 语义缓存 · 评估框架(recall@k、MRR、nDCG)— 计划在 openspec/specs/mnemos-differentiators/ 中。
学术
综合项目 · 教授 Christian Pérez · 学生:Gat · 2026
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables semantic search over an Obsidian vault using natural language, retrieving relevant notes and extracted conclusions.MIT
- AlicenseNot gradedqualityAmaintenanceProvides a RAG system for your Obsidian Vault using DuckDB and vector search, enabling semantic search and note retrieval through AI agents.2MIT
- FlicenseAqualityDmaintenanceEnables indexing local documents (PDF, Markdown, text, code) into a knowledge base and querying them via semantic search using local embeddings, all running privately on your machine.4
- AlicenseNot gradedqualityAmaintenanceEnables semantic search and retrieval over an Obsidian vault using local or API-based embeddings, allowing AI assistants to find notes by meaning, get related content, and pull context during conversations.13MIT
Related MCP Connectors
Search and reason over your Obsidian-style Markdown vault, right from ChatGPT.
Connect AI assistants to your GitHub-hosted Obsidian vault to seamlessly access, search, and analy…
Search your Obsidian vault to quickly find notes by title or keyword, summarize related content, a…
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Gatoco/mnemos-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server