EdgeEver MCP server
Commit Nov 8, 2026An AI-native knowledge base that allows AI agents to read and write notes via native MCP support.
Also known as scupro/edgeever
- Stars
- 6
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- —
- Last commit
- Nov 8, 2026
- License
- AGPL-3.0
What is EdgeEver MCP?
EdgeEver is an AI-native knowledge base that allows AI agents to read and write notes via native MCP support.
Summary generated from the project's README
About EdgeEver MCP server
Managing a personal knowledge base that AI agents can interact with is what EdgeEver enables. This open-source workspace serves as an alternative to Evernote or Obsidian, providing a structured three-pane productivity workflow with a tree view, note list, and editor. It is designed for users who need a reliable, low-latency system for capturing inspiration and compounding knowledge across multiple devices.
Developers and researchers can deploy the service for free using Cloudflare's serverless tier or host it themselves via Docker on a VPS, NAS, or home server. The system supports native MCP protocols, allowing external AI agents to read and write notes directly to your database. It also features a companion sidebar AI assistant and supports Bring Your Own Key (BYOK) for OpenAI, Claude, and Gemini. Users can manage their data through desktop clients for macOS, Windows, and Linux, or via an Android mobile app.
What you can do with it
- 1Automated note management by agentsAn AI agent can directly read from or write new entries into your knowledge base using native MCP support.
- 2Zero-cost serverless deploymentYou can run your entire knowledge base on Cloudflare's free tier without needing to maintain a dedicated server.
- 3Web content archivingUse the browser extension to extract clean Markdown from articles or specific social media posts.
Written from the project's README. Every use case is checked against a quote from it.
Automated checks
Ran Nov 15, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablescupro/edgeever is public and reachablegithub.com/scupro/edgeever ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)AGPL-3.0spdx.org/licenses/AGPL-3.0.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-08github.com/scupro/edgeever ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeNot checked
README
View on GitHubEdgeEver is a modern, open-source notes and knowledge base workspace. It preserves the classic, high-efficiency three-pane productivity workflow (Tree View · Note List · Editor) while offering an open data architecture and seamless AI Agent integration for complete ownership and smart productivity.
💡 Serverless & 100% Free Forever EdgeEver can run within Cloudflare's free quotas with no server purchase or VPS maintenance. Users who prefer a VPS, NAS, or home server can deploy the same application with Docker.
⭐ If EdgeEver is useful to you, consider giving it a Star. Your support helps more people discover the project.
Why EdgeEver
Many users of note-taking apps and personal knowledge bases simply want a reliable, open, and fast personal knowledge base. However, existing mainstream solutions all present tradeoffs:
- Evernote: Offers the smoothest, most seamless multi-device cloud sync experience. Yet as it evolved into a full-featured commercial suite, ads and add-ons multiplied while system overhead climbed. Meanwhile, free-tier quotas tightened, premium AI subscriptions grew expensive, and it remains difficult to integrate into modern self-hosted and private AI workflows.
- Obsidian: Open files, closed-source core. Official Sync is paid and third-party sync is tedious; relying entirely on flat local file scanning causes noticeable cold-start and search lag once notes reach thousands or heavy plugins are loaded; storing images and attachments alongside notes quickly bloats vaults, making mobile sync sluggish and leaving orphaned files behind; and it is overly heavy for lightweight, capture-anywhere use.
- Memos & Stream Notes: Clean and simple, but their social-timeline layouts differ fundamentally from the structured productivity of a classic three-pane workflow.
- SiYuan & Block-based PKMs: Powerful with self-hosting support, but their granular "block-level" architecture imposes noticeable cognitive overhead for quick daily capture and continuous prose writing. Furthermore, they lack a true zero-cost serverless deployment tier, and multi-device sync relies on paid official subscriptions or paying extra to unlock S3/WebDAV sync features with your own storage.
EdgeEver fills this gap: The entire stack is open source, delivering truly seamless, low-latency, instant multi-device cloud sync with full self-hosting freedom. It keeps the three-pane layout you know, stays silky-smooth and lightweight even with 10,000+ notes, and ships native AI agents with zero-cost deployment.
All-Platform Clients & Browser Extension
Client Downloads
Browser Web Clipper Extension
- Extract clean Markdown from web articles or clip selected text and images directly to your self-hosted instance.
- Deep structured clipping for 20+ platforms: social (X, Xiaohongshu, WeChat, Jike, Douban, Threads, Bluesky), dev & reading communities (Zhihu, Reddit, GitHub, V2EX, Hacker News, Medium, Substack), and media/podcasts (Bilibili, YouTube, Xiaoyuzhou).
Feature Matrix & Capabilities
An all-in-one workspace connecting inspiration capture, deep creation, and knowledge compounding:
| Module | Core Features | Advanced Features |
|---|---|---|
| Writing & Editing | • Classic 3-pane layout with resizable panel width• Seamless toggle between rich-text WYSIWYG & raw Markdown• Syntax highlighting, math formulas, task lists & Focus Mode• Second-level auto-save, instant multi-device sync & revision history | • Native Poster Notes: Canvas-based structured visual card layouts, double-click to edit• WeChat Official Account Formatter: One-click copy with inline CSS to newsletters/blogs• Poster Image Export: Curated themes, Monogram avatars; export to MD, HTML, or PDF |
| Organization & Search | • Infinite-depth notebooks & folder tree navigation• Tagging system, drag-and-drop sorting, batch note merging/moving• Millisecond-level full-text search with historical filtering | • Standalone Knowledge Base Publishing: One-click publish entire notebooks as static documentation sites with tree navigation & password protection• Multi-user spaces & physical tenant data isolation |
| Capture & Media | • Universal attachment uploads (PDF, Office, audio/video, archives)• Browser extension precision clipping for web text & images• Mobile system share sheet to save photos & articles | • Deep Multi-Platform Structured Clipping: Deeply adapted for Xiaohongshu, X (Twitter), Zhihu, Reddit, GitHub, YouTube, Bilibili, and 20+ sources• WeChat Chat History Archiving: One-click full transcript import on macOS with timestamps, quotes, emojis, and media• Client-Side Image Compression: Automatic 50%–90% silent size reduction; chunked streaming safely supports up to 1 GiB files |
| Data & Visuals | • Native table editing & Markdown table rendering• Native Mermaid diagrams (flowcharts, sequence diagrams, architecture) | • AI Multi-Dimensional Tables: Prompt-to-database with automatic field typing (tags, dates, selects)• Public Online Collection Forms: One-click shareable forms without login, syncing responses into tables in real time• AI Diagram & Infographic Generation: In-place creation of timelines, comparison charts, quadrant matrices & mind maps |
| AI & Ecosystem | • Companion sidebar AI assistant for summaries, proofreading & translations• BYOK (Bring Your Own Key): compatible with OpenAI, Claude, Gemini & custom endpoints | • Native Agent Protocols (MCP & ACP): Built-in MCP for external agent read/write; ACP in desktop app to drive local coding agents directly• AI RSS Daily Digest: Automated feed aggregation, AI noise filtering, and structured digest generation• Extensible Plugin Development API |
| Multi-Device Sync & Foundation | • Seamless, low-latency sync across desktop, mobile, and web with zero device limits• Official client support across Web, macOS, Windows, Linux, and Android• Offline-first with auto-sync queue: write smoothly offline, automatically resume and sync when reconnected• Standard SQLite database, lossless ZIP backup & Markdown export/import | • Zero Server Cost for Life: 100% runnable on Cloudflare Free tier (~150k short notes & ~50k images)• Flexible Docker Self-Hosting: One-click deployment for VPS, NAS, or home server with millions of notes & instant sync• Server-side brute-force protection and tenant security |
👉 Explore all 18 step-by-step showcases with screenshots: Complete Showcase & Workflows Guide
Deployment
Cloudflare is the recommended zero-server deployment. Docker is available for users who prefer a VPS, NAS, or home server.
For Cloudflare, choose either of the following online deployment options:
Option A: Deploy with an AI Agent (Recommended)
Copy the prompt below directly into an AI Agent (such as Codex, Claude, Cursor, WorkBuddy, Antigravity, OpenClaw, Hermes Agent, etc.). During execution, if access to GitHub or Cloudflare is required, review the requested permissions and follow the prompts to authorize access.
Deploy EdgeEver entirely through GitHub and Cloudflare:
1. Fork https://github.com/scupro/edgeever.
2. Create D1 `edgeever` in Cloudflare and prepare an S3-compatible object storage bucket for attachments.
3. In Workers & Pages, create a Worker named `edgeever` from the Fork's `main` branch.
Use the repository root, keep Cloudflare's default Workers Builds deploy command,
and ensure its API token can read and edit D1. Select Save and Deploy.
4. After the Worker is created, add the user's chosen password as the runtime Secret
`AUTH_PASSWORD` (preferably at least 32 characters).
The username defaults to `admin`.
If the user specifies another, set `AUTH_USER` as a Workers Builds
variable before the next build.
5. Run the build again, verify `/api/health` and `/api/openapi.json`, then log in
with that administrator username and password.
6. Optional: to take future releases, merge an upstream tag into the Fork and push
(`git fetch upstream --tags && git merge upstream/<tag>`); the repository ships no
scheduled upstream-sync workflow.
Detailed requirements: AI Agent Cloudflare Deployment.
Option B: Manual Online Deployment
Complete setup in 6 web steps:
- Fork the Repository: Click Fork at the top right of GitHub to fork EdgeEver into your personal account.
- Create Cloudflare Resources: Create D1
edgeeverand prepare an S3-compatible object storage bucket (Backblaze B2, Tencent COS, Alibaba OSS, MinIO or Wasabi). - Import & Configure the Project: Create a Worker named
edgeeverfrom the Fork'smainbranch in Cloudflare Workers & Pages. Use the repository root and keep Cloudflare's default Workers Builds deploy command, which runs in Cloudflare. Ensure its API token can read and edit D1. The deploy command creates the bindings; you may still adjustwrangler.toml(including a[vars]block for non-secret variables). - Choose the Administrator Password: Choose an administrator password, preferably at least 32 characters. Once the Worker is created, save it as the runtime Secret
AUTH_PASSWORD. - Build & Verify: Save and Deploy creates the Worker and starts a build. The initial build primarily initializes the Worker instance; if it fails due to missing administrator Secret, this is expected behavior. Simply add the runtime Secret
AUTH_PASSWORDfrom Step 4 and click retry. The username defaults toadmin; to use another, setAUTH_USERas a Workers Builds variable before retrying. Once deployed, confirm/api/healthreturns200, then log in with the configured username and password. - Optional: Keep the Deployment Updated: The Fork ships no scheduled upstream-sync workflow. To take a new release, merge the upstream tag into
main(git fetch upstream --tags && git merge upstream/<tag>) and push; Cloudflare rebuilds automatically.
📖 For full step-by-step instructions and configuration details, see the Online Deployment Guide.
Custom Domain & Access: After deployment, you can directly use the default
*.workers.devdomain assigned by Cloudflare, or attach your own custom domain in the Worker settings under Settings → Domains & Routes.Object Storage: EdgeEver keeps attachments and images in an S3-compatible object storage service that you provide; it binds no Cloudflare R2 bucket. Backblaze B2 offers a free 10 GB tier, and Tencent COS, Alibaba OSS, MinIO and Wasabi work as well. Set the
EDGE_EVER_S3_*deployment variables before the first build.
Option C: Docker on a VPS or NAS
Docker runs the same frontend, API routes, services, authentication, MCP implementation, and migrations as Cloudflare. The container uses SQLite with local files or S3-compatible attachment storage and supports amd64 and arm64. See Deploy EdgeEver with Docker and Self-hosting and Docker architecture.
Pull the published image and start it directly:
docker pull ghcr.io/scupro/edgeever:latest
docker run -d --name edgeever -p 8787:8787 \
-e AUTH_USER=admin \
-e AUTH_PASSWORD='choose-a-strong-password' \
-v edgeever-data:/data \
ghcr.io/scupro/edgeever:latest
Then open http://localhost:8787 and sign in with the credentials above. To update, docker pull ghcr.io/scupro/edgeever:latest, remove the container, and run the same docker run again; the edgeever-data volume keeps your notes.
Until the first release is published, pull ghcr.io/scupro/edgeever:main instead of :latest; publishing a vX.Y.Z release publishes that tag and moves :latest.
See the Docker deployment guide for Compose, TLS, backups and configuration.
Advanced Configuration & Management
Data Backup, Export & Migration
Whether for routine data backups, seamless cross-instance migration, or switching from existing note platforms, EdgeEver provides reliable, complete data mobility:
- EdgeEver Native Backup & Restore: In the app, go to Profile -> Import and export to export or import an EdgeEver ZIP archive at any time. The extracted
notes/directory keeps clean plaintext Markdown files and attachments for local archival or direct reading, while internal structured metadata enables lossless full restoration across different EdgeEver instances. Importing preserves existing unrelated data on the target instance and automatically overwrites records with matching IDs. - Smooth Migration from Third-Party Apps:
- Evernote Migration: Please refer to docs/evernote-migration-guide.md
- flomo Migration: Please refer to docs/flomo-migration-guide.md
- Memos Migration: Please refer to docs/memos-migration-guide.md
- Notion Migration: Please refer to docs/notion-migration-guide.md
AI & MCP Protocol Integration
Create an API token in Profile -> API / MCP and copy the Remote MCP configuration in one click to let AI Agents such as Claude Code, Cursor, Antigravity, and OpenClaw securely manage your knowledge base within account permissions. EdgeEver supports full CRUD for text notes, visual diagram notes (mind maps, flowcharts, and architecture diagrams), structured table notes, and native poster notes. Agents can also manage notebook hierarchies, tags, attachments, revision history, note templates, and AI instructions.
💡 Inspiration: Make AI your true knowledge orchestrator and creative co-pilot—instantly turn concepts into interactive mind maps, flowcharts, architecture diagrams, and structured tables, while supplying private context to your AI Agents. Paired with EdgeEver’s powerful rich-text editing and elegant typography, AI-assisted content becomes beautifully structured, polished, and publication-ready knowledge assets.
Multi-Account Login & Storage Management
Once deployed, a single instance supports multi-account login.
The instance administrator can create, disable, or reset member accounts in Profile -> User accounts. Each member gets a fully isolated personal workspace, including notebooks, notes, attachments, Trash, import/export, and MCP tokens. An intuitive storage usage dashboard provides real-time visibility into attachment storage consumed across the whole instance as well as per account.
Storage Extension & Third-Party Image Hosting
- OSS Object Storage: The instance owner can configure S3-compatible object storage under Settings → Advanced → OSS object storage. Changing storage does not migrate or affect existing attachments.
- Desktop PicGo Integration: The desktop app optionally integrates with a local PicGo client service, uploading inserted images directly to GitHub, SM.MS, Upyun, Qiniu, or custom image hosts to minimize self-hosted storage and bandwidth consumption.
Plugins and Themes
Web and desktop apps support functional plugins and custom themes, installable from the official marketplace, GitHub, or a Manifest URL, with seamless sync across your workspace. Developers can extend capabilities using @edgeever/plugin-api; see the plugin development guide and marketplace submission policy.
Developer Guide & Contributing
Quick Start
bun install
bun run dev
Local development signs in automatically; fresh databases use owner / edgeever-local-dev. Log out to test the login screen.
Architecture & Tech Stack
Tech Stack
- Bun workspace monorepo with Web, API, official site, and shared type package.
- Frontend: Vite, React, React Router, TanStack Query, Tailwind CSS, shadcn/ui, and Radix UI.
- Editor: TipTap / ProseMirror with Markdown support; PWA uses vite-plugin-pwa, Workbox, and Dexie.
- Android app: Expo + React Native in
apps/mobile, with SQLite local storage and incremental sync. - Native desktop app: Electron + Rust sidecar combines a consistent cross-platform experience with high-performance local data services; SQLite enables offline editing, incremental sync when back online, and local backups.
- Web clipper: Manifest V3, Mozilla Readability, and Turndown for Chrome, Microsoft Edge, and Firefox.
- Backend: one Hono/Zod business application with REST API and Remote MCP; Cloudflare uses Workers/D1 plus external S3-compatible object storage, while Docker uses Bun/SQLite/local files or S3.
- Official site: Astro static site in
apps/site, deployable to Cloudflare Pages.
Project Structure
apps/web Vite + React frontend, PWA, offline drafts, and sync queue
apps/extension Chrome/Edge/Firefox Manifest V3 web clipper
apps/api Cloudflare Worker + Hono API, MCP endpoint
apps/mobile Expo + React Native Android app
apps/desktop Electron desktop shell, preload bridge, and native packaging
apps/site Astro official website, deployable independently
packages/client Shared API client for web and mobile apps
packages/shared Shared types, Zod schemas, TipTap / Markdown conversion
crates/desktop-sidecar
Rust sidecar for local SQLite, offline data, backups, and resources
scripts Wrangler wrapper, password hash, CLI, MCP stdio bridge, Evernote ENEX import
migrations Shared append-only D1/SQLite database migrations
docs Architecture, migration, and deployment docs
.github/workflows CI for web, mobile, desktop packaging, deployment, and releases
wrangler.toml Cloudflare Workers, Assets, D1 configuration
Data Storage Model
EdgeEver persists note content simultaneously across three formats to balance editing performance, open portability, and full-text search speed:
content_json TipTap/ProseMirror document, the editor source of truth
content_markdown API, Agent, export, and external ecosystem interoperability
content_text Full-text search, summary generation, and indexing
Core Mechanisms & Design Decisions
- Client-Side Silent Compression: To avoid consuming compute quota on serverless runtimes like Cloudflare Workers, image compression occurs in the Web client before upload (converting PNG/JPEG/WebP/AVIF to WebP when beneficial, max edge 2560px; original retained if compression does not reduce size). REST API and MCP endpoints store client-supplied binaries directly without additional server-side transcoding.
- Offline-First & Sync Timing: Web, PWA, and desktop clients automatically upload memo edits after 30 seconds of typing inactivity, and poll for remote changes every 5 minutes while visible; window refocus and manual refresh remain immediate. Configurable via
DEFERRED_MEMO_SYNC_DELAY_MSandBACKGROUND_WORKSPACE_REFRESH_INTERVAL_MSinapps/web/src/lib/workspace-refresh.ts.
EdgeEver MCP questions
Can I host EdgeEver for free?
Yes, EdgeEver can run within Cloudflare's free quotas without needing a VPS or server maintenance.
Does EdgeEver support MCP?
Yes, it includes built-in MCP support for external agent read/write capabilities.
How can I self-host EdgeEver?
You can deploy EdgeEver using Docker on a VPS, NAS, or home server.
What kind of AI models can I use?
It supports Bring Your Own Key (BYOK) for OpenAI, Claude, Gemini, and custom endpoints.