tianma-if/edgeever

Open-source, AI-native knowledge base & Evernote alternative with native MCP. Zero-cost on Cloudflare or Docker

Summary

EdgeEver is a self-hosted, open-source notes and knowledge-base application that revives Evernote's three-pane layout while adding native Model Context Protocol (MCP) support for AI agents. It runs on either Cloudflare's free tier (Workers + D1 + R2) or Docker, with SQLite as the underlying data store so notes remain portable. The codebase is a TypeScript Bun monorepo covering web, mobile, desktop, API, and a browser clipper.

What it is useful for

Engineers who want a personal or team knowledge base they fully own — without Evernote's bloat or vendor lock-in. The MCP endpoint lets AI tools such as Claude Code, Cursor, Codex, and Antigravity read and write notes, manage notebooks and tags, and operate on diagram and table notes through standardized tooling. A REST API, plugin system, and ZIP-based export (Markdown plus front matter, nested folders, and attachments) make it usable for scripting, automation, and clean data extraction at any time.

The classic three-pane workflow fits engineers who prefer structured navigation over block-based or graph views, and the offline-first PWA plus Electron desktop client keep it usable without connectivity. A Manifest V3 web clipper for Chrome, Edge, and Firefox extracts pages from sources including GitHub, Reddit, X, Zhihu, and Xiaohongshu into clean Markdown directly to a self-hosted instance.

How engineers can use it

For local development, the monorepo boots with Bun using bun install followed by bun run dev. Fresh local databases auto-sign-in; the default credentials documented in the README are owner / edgeever-local-dev, and you should log out to see the login screen.

For production, two paths are documented:

• Cloudflare: fork the repo, create a D1 database named edgeever and an R2 bucket named edgeever-resources, deploy as a Worker from the fork's main branch, and set the EDGE_EVER_AUTH_PASSWORD runtime secret (the README recommends at least 32 characters). A copy-paste prompt is provided so an AI agent can drive the entire deployment. Note that R2 requires activating a subscription and adding a payment method even though workloads stay within the free allowance.
• Docker on a VPS or NAS: a single installer command (curl -fsSL https://edgeever.org/install.sh | bash) pulls the GHCR image and starts the container; daily auto-updates are configured by default, with manual updates via ~/edgeever/update.sh.

To wire AI tools into the data, open Profile → API/MCP, generate an API token, and copy the one-click Remote MCP configuration into a compatible client. Engineers extending the system can use the @edgeever/plugin-api package; plugin authoring is documented in docs/plugin-development.md.

A few documented caveats: the iOS build requires an Apple ID from outside mainland China, and advanced sections such as S3-compatible OSS object storage are referenced but partially truncated in the README, so readers should consult the linked deployment and best-practices docs for full configuration details.