NornWeave

NornWeave

Open-source, self-hosted Inbox-as-a-Service API for AI agents. It enables agents to manage email inboxes, send/receive emails, search messages, and wait for replies via REST or MCP.

Category
访问服务器

README

<p align="center"> <img src="web/static/images/Nornorna_spinner.jpg" alt="The Norns weaving fate at Yggdrasil" width="400"> </p>

<h1 align="center">NornWeave</h1>

<p align="center"> <em>"Laws they made there, and life allotted / To the sons of men, and set their fates."</em><br>

  • Voluspa (The Prophecy of the Seeress), Poetic Edda, Stanza 20 </p>

<p align="center"> <strong>Open-source, self-hosted Inbox-as-a-Service API for AI Agents</strong> </p>

<p align="center"> <a href="https://github.com/DataCovey/nornweave/actions"><img src="https://github.com/DataCovey/nornweave/workflows/CI/badge.svg" alt="CI Status"></a> <a href="https://github.com/DataCovey/nornweave/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-Apache--2.0-blue.svg" alt="License"></a> </p>


What is NornWeave?

Standard email APIs are stateless and built for transactional sending. NornWeave adds a stateful layer (Inboxes, Threads, History) and an intelligent layer (Markdown parsing, Semantic Search) to make email consumable by LLMs via REST or MCP.

In Norse mythology, the Norns (Urdr, Verdandi, and Skuld) dwell at the base of Yggdrasil, the World Tree. They weave the tapestry of fate for all beings. Similarly, NornWeave:

  • Takes raw "water" (incoming email data streams)
  • Weaves disconnected messages into coherent Threads (the Tapestry)
  • Nourishes AI Agents with clean, structured context

Features

Foundation (The Mail Proxy)

  • Virtual Inboxes: Create email addresses for your AI agents
  • Webhook Ingestion: Receive emails from Mailgun, SES, SendGrid, Resend
  • IMAP/SMTP: Poll existing mailboxes (IMAP) and send via SMTP for any provider or self-hosted server
  • Persistent Storage: SQLite (default) or PostgreSQL with abstracted storage adapters
  • Email Sending: Send replies through your configured provider

Intelligence (The Agent Layer)

  • Content Parsing: HTML to clean Markdown, cruft removal
  • Threading: Automatic conversation grouping via email headers
  • Thread Summarization: LLM-generated thread summaries (OpenAI, Anthropic, Gemini) for list views and token savings
  • MCP Server: Connect directly to Claude, Cursor, and other MCP clients
  • Attachment Processing: Extract text from PDFs and documents

Enterprise (The Platform Layer)

  • Semantic Search: Vector embeddings with pgvector
  • Real-time Webhooks: Get notified of new messages
  • Rate Limiting: Protect against runaway agents
  • Multi-Tenancy: Organizations and projects

Quick Start

Try NornWeave with no config: demo mode gives you a local sandbox and a pre-configured inbox.

pip install nornweave[mcp]
nornweave api --demo

API: http://localhost:8000. Open /docs, call GET /v1/inboxes to get the demo inbox id, then use MCP (nornweave mcp) to send and list messages. When you're ready for real email, set a provider and domain (see below).

Install options

# Base (SQLite, Mailgun/SES/SendGrid/Resend)
pip install nornweave

# With IMAP/SMTP support
pip install nornweave[smtpimap]

# With PostgreSQL
pip install nornweave[postgres]

# With MCP server (recommended for agents)
pip install nornweave[mcp]

# Full
pip install nornweave[all]

Real email (after demo)

Create a .env file and set your provider and domain:

# .env — for inbox creation with a real provider
EMAIL_PROVIDER=mailgun   # or ses, sendgrid, resend, imap-smtp
EMAIL_DOMAIN=mail.yourdomain.com
# ... plus provider-specific keys (see .env.example)

See Configuration for all settings.

nornweave api

Data is stored in ./nornweave.db (SQLite default).

Using Docker (Recommended for Production)

# Clone the repository
git clone https://github.com/DataCovey/nornweave.git
cd nornweave

# Copy environment configuration and set EMAIL_DOMAIN + provider keys (see above)
cp .env.example .env

# Start the stack
docker compose up -d

# Run migrations (PostgreSQL only — SQLite tables are created automatically)
docker compose exec api alembic upgrade head

Using uv (Development)

# Clone the repository
git clone https://github.com/DataCovey/nornweave.git
cd nornweave

# Install dependencies
make install-dev

# Copy environment configuration and set EMAIL_DOMAIN + provider keys (see above)
cp .env.example .env

# Run migrations (PostgreSQL only — SQLite tables are created automatically)
# make migrate

# Start the development server
make dev

API Overview

Create an Inbox

curl -X POST http://localhost:8000/v1/inboxes \
  -H "Content-Type: application/json" \
  -d '{"name": "Support Agent", "email_username": "support"}'

Read a Thread

curl http://localhost:8000/v1/threads/th_123

Response (LLM-optimized):

{
  "id": "th_123",
  "subject": "Re: Pricing Question",
  "messages": [
    { "role": "user", "author": "bob@gmail.com", "content": "How much is it?", "timestamp": "..." },
    { "role": "assistant", "author": "agent@myco.com", "content": "$20/mo", "timestamp": "..." }
  ]
}

Send a Reply

curl -X POST http://localhost:8000/v1/messages \
  -H "Content-Type: application/json" \
  -d '{
    "inbox_id": "ibx_555",
    "reply_to_thread_id": "th_123",
    "to": ["client@gmail.com"],
    "subject": "Re: Pricing Question",
    "body": "Thanks for your interest! Our pricing starts at $20/mo."
  }'

MCP Integration

NornWeave exposes an MCP server for direct integration with Claude, Cursor, and other MCP-compatible clients.

Available Tools

Tool Description
create_inbox Provision a new email address
send_email Send an email (auto-converts Markdown to HTML)
search_email Find relevant messages in your inbox
wait_for_reply Block until a reply arrives (experimental)

Configure in Cursor/Claude

pip install nornweave[mcp]
{
  "mcpServers": {
    "nornweave": {
      "command": "nornweave",
      "args": ["mcp"],
      "env": {
        "NORNWEAVE_API_URL": "http://localhost:8000"
      }
    }
  }
}

Architecture

NornWeave uses a thematic architecture inspired by Norse mythology:

Component Name Purpose
Storage Layer Urdr (The Well) Database adapters (PostgreSQL, SQLite), migrations
Ingestion Engine Verdandi (The Loom) Webhook + IMAP ingestion, HTML→Markdown, threading, LLM thread summarization
Outbound Skuld (The Prophecy) Email sending, rate limiting, webhooks
Gateway Yggdrasil FastAPI routes, middleware, API endpoints
MCP Huginn & Muninn Read resources and write tools for AI agents

Supported Providers

Provider Sending Receiving Auto-Route Setup
Mailgun yes yes yes
AWS SES yes yes manual
SendGrid yes yes yes
Resend yes yes yes
IMAP/SMTP yes (SMTP) yes (IMAP polling) config

Documentation

Repository Structure

This is a monorepo:

Directory Purpose
src/nornweave/ Main Python package: adapters (Mailgun, SES, SendGrid, Resend, SMTP/IMAP), core config, models, huginn (MCP resources), muninn (MCP tools), skuld (outbound/sending), urdr (storage, migrations), verdandi (ingestion, parsing, threading), yggdrasil (FastAPI gateway), search, storage backends
clients/python/ Python client SDK (nornweave-client)
packages/n8n-nodes-nornweave/ n8n community node for NornWeave
tests/ Test suite: fixtures/, integration/, unit/, e2e/
web/ Hugo documentation site (content/docs/)
scripts/ DB init, migrations, dev setup
skills/ Distributable AI assistant skills (e.g. nornweave-api)
openspec/ Specs (specs/) and change artifacts (changes/, changes/archive/)

Contributing

We welcome contributions! Please see our Contributing Guide for details.

License

NornWeave is open-source software licensed under the Apache 2.0 License.


<p align="center"> <sub>Image: "Nornorna spinner odet tradar vid Yggdrasil" by L. B. Hansen</sub><br> <sub><a href="https://commons.wikimedia.org/w/index.php?curid=164065">Public Domain</a></sub> </p>

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选