FlareMo

FlareMo

Zero Servers · Zero Upkeep · 24/7 Global Edge Uptime · Absolute Data Ownership

For individuals: a quiet, focused thought capture space and second brain. For teams: a shared knowledge base with fine-grained roles.

English • 简体中文 • 日本語 • Français • Español • 한국어 • Русский • العربية

Why FlareMo?

Tools like Flomo and Memos proved the immense value of low-friction memo capture and a distraction-free timeline. However, self-hosting a traditional note-taking setup typically means paying for a VPS, configuring Docker and PostgreSQL, scripting automated backups, and dreading disk or hardware failures.

FlareMo answers a simpler question: Can you get a 24/7 online, resilient, globally accelerated knowledge base with zero server maintenance, using just a free Cloudflare account?

• Truly Serverless: Both code and static assets run on Cloudflare Workers edge nodes near you with millisecond latency.
• Enterprise-grade durability out of the box: Cloudflare D1 handles notes and metadata; Cloudflare R2 stores media attachments with multi-region replication.
• AI-Native Second Brain: Built-in MCP endpoints and Agent Memory hub allow AI agents (Claude, Cursor, Codex, ChatGPT) to read and update your long-term preferences and memory scopes.
• Quiet for one, powerful for many: Default is an encrypted, private single-user sanctuary. Enable team mode, and it instantly transforms into a collaborative workspace with roles and three-tier visibility.
• Minimal, not simplistic: The interface stays quiet and every control earns its place — nothing decorative shouting for attention, nothing useful missing.

✨ Key Features

1. Instant Capture & Inspiring Review

• Capture in milliseconds: Card-style timeline, tags, Markdown/GFM, and previews for image and audio attachments.
• Lightning-fast search: SQLite FTS5 full-text indexing with query operators (has:attachment, is:pinned, before:YYYY-MM-DD, after:YYYY-MM-DD, in:timeline|archive|trash).
• Semantic "Find" (Vector Search): Workers AI embeddings paired with Vectorize derived vector indexes for contextual recall; re-verifies permissions against D1 and seamlessly falls back to FTS5.
• Thought activation: Built-in Daily Review (on this day), Random Walk (wandering through tag and backlink graphs with postcard summaries), and related note recommendations.
• Revision history: Full version diffs and one-click historical restore.

2. AI Long-term Memory & Native MCP

• Agent Memory: Through /memory/mcp, AI agents can record and update cross-session long-term memory (preferences, project decisions, constraints, lessons).
• Human in the loop: Review, verify, lock, or correct AI-recorded memories at /memory.
• Open ecosystem: Standard /mcp (Streamable HTTP MCP) endpoint to query and append notes programmatically.

3. Projects & Tasks

• Group work under projects: Organize notes and to-dos into projects, with a kanban board (drag between status columns), priorities, manual sort order, and due dates.
• Personal by design, reversible deletion: Tasks belong to a single owner; deleting moves them to a recycle bin until restored or automatically purged.

4. Calendar

• Tasks are the schedule's source of truth: The /calendar month view fills past days with what you wrote and future days with what's due — drag to reschedule, quick-add dated tasks, or scan the agenda list.
• Overdue reminders: Overdue tasks raise in-app notifications, with optional browser Web Push.

5. Team Collaboration & 3-Tier Visibility

• Role governance: owner, admin, and member roles. Admins invite members via one-time activation links (members choose their own passwords; admins never handle plaintext credentials).
• 3-tier visibility:
  • Private: Only author can view.
  • Team: Shared read-only with active team members.
  • Public: Anonymous read-only via time-limited share links.
• Safe offboarding: Removing a member triggers reliable background cleanup that purges private data while preserving team and public notes.

6. Offline First & PWA Experience

• Installable PWA: Install to macOS, Windows, iOS, or Android home screen with native feel.
• Reliable offline sync: Drafts save locally instantly. Offline submissions and uploads queue up and replay automatically when connectivity is restored.
• Live voice capture: Access /capture for real-time streaming speech-to-text (ASR) transcription.

7. Secure Better Auth Application Security

• Better Auth powered: HttpOnly, SameSite=Lax browser cookie sessions; revocable memos_pat_ Personal Access Tokens for scripts, CLI, and MCP.
• Strict Origin protection: State-changing requests enforce exact origin whitelisting. Cloudflare Access remains available as an optional outer defensive perimeter.

8. Memos Compatibility & Seamless Migration

• Memos /api/v1 compatibility: Provides core Memos API endpoints (camelCase default, legacy snake_case via header) and OpenAPI schema.
• Third-party apps ready: Works directly with mobile clients like Moe Memos.
• Bi-directional import & export: One-click import from Memos / flomo with conflict strategies and full raw export bundles.

How Generous Is Cloudflare's Free Tier?

Many assume "free" means "severely limited". For text-heavy personal knowledge bases, Cloudflare's free quota is virtually inexhaustible:

Comparison: Cloudflare Native vs Home NAS vs Traditional VPS

5-Minute Quick Deployment

Method 1: One-click Deploy to Cloudflare

Clones the repository into your GitHub account and provisions D1, R2, Queues, and Vectorize automatically. After the initial deploy, set FLAREMO_PUBLIC_URL and secrets (see docs/en/deploy.md). If the first attempt reports "Github API Limit Exceeded", wait a few minutes and retry.

Method 2: GitHub Action (self-hosted fork)

On your fork, run Deploy to Cloudflare from Actions to provision resources, publish the Worker, and sync auth secrets. Pushes do not publish. See docs/github-action-deploy.md or the English guide.

Method 3: Deploy with an AI Agent (Recommended)

Give the repository to an agent capable of executing terminal commands (e.g. Claude Code, Cursor Agent, Codex) along with docs/agent-deploy.md:

"Please deploy FlareMo to my Cloudflare account following docs/agent-deploy.md."

Method 4: Manual 3-Step Deployment

1. Create Cloudflare Resources

pnpm exec wrangler whoami
pnpm exec wrangler d1 create flaremo
pnpm exec wrangler r2 bucket create flaremo-attachments

Or run pnpm provision:remote instead: it creates the missing D1 / R2 / Queue / Vectorize resources and writes the D1 database_id into wrangler.jsonc for you. It is idempotent — existing resources are skipped.

2. Configure Settings & Secrets

cp wrangler.jsonc.example wrangler.jsonc

Fill in the generated database_id and set FLAREMO_PUBLIC_URL to your production domain. Then set secrets:

pnpm exec wrangler secret put BETTER_AUTH_SECRET --config ./wrangler.jsonc
pnpm exec wrangler secret put FLAREMO_BOOTSTRAP_SECRET --config ./wrangler.jsonc

3. Deploy

pnpm deploy:dry-run
pnpm deploy

(The full pnpm verify gate runs only when the maintainer explicitly asks for it.) Visit your production domain at /setup and enter the FLAREMO_BOOTSTRAP_SECRET to initialize your Owner account.

Detailed guides: Deployment Guide · GitHub Action deploy · Update Guide.

Architecture & Tech Stack

flowchart LR
  Browser["FlareMo Web UI (React 19 / PWA)"] --> Worker["Cloudflare Worker"]
  Clients["Memos Clients / Scripts / MCP"] --> Worker

  Worker --> Auth["Better Auth (Session / PAT)"]
  Worker --> D1["Cloudflare D1 (Memos / Relations / Settings)"]
  Access["Cloudflare Access (Optional Outer Perimeter)"] -.-> Worker
  Worker --> R2["Cloudflare R2 (Attachments & Exports)"]
  Worker --> Assets["Workers Static Assets"]

• Runtime: Cloudflare Workers
• Frontend: React 19, Vite, TanStack Router, Tailwind CSS 4, Radix UI
• Database: Cloudflare D1, Drizzle ORM
• Storage: Cloudflare R2
• Auth: Better Auth (HttpOnly cookie session + revocable memos_pat_)
• AI & Search: Workers AI, Vectorize, SQLite FTS5

Star History

License

Open-sourced under the GNU AGPL-3.0 license. Copyright (c) 2026 realchendahuang.