io.github.KuvopLLC/better-bear MCP Server
io.github.KuvopLLC/better-bear
Read, search, create, and manage Bear notes via CloudKit with AI-powered context library support.
What this kind of tool is for
- Create, search, or update items in a notes, docs, or tasks service from your agent.
- Turn a meeting transcript or a long document into a structured summary.
- Keep a daily log or standup digest without leaving the terminal.
How to compare servers like this
- Service scope:
- Confirm the tool covers the exact service and surfaces (pages, tasks, calendars) you need.
- Auth model:
- See whether it uses an API token you control and what permissions that token grants.
- Read vs write:
- Start with read-only workflows; give write access only after the first results look right.
- Maintenance signal:
- For anything tied to a third-party API, prefer tools that track that API's changes.
This page carries your copy-paste config: pick your MCP client in the install section on this page and use the copy button.
What is the io.github.KuvopLLC/better-bear MCP server?
The better-bear MCP server provides full access to Bear notes through CloudKit, enabling AI agents to read, search, create, edit, tag, and manage notes. It includes a context library feature that syncs a curated subset of notes optimized for LLM consumption, following Karpathy's knowledge base pattern.
better-bear connects Claude and other AI agents to your Bear notes library. You can search, create, edit, and organize notes directly from Claude, plus set up a context library—a synced folder of tagged notes that Claude can efficiently query without loading your entire note collection. It works via CloudKit authentication and includes both CLI and MCP server interfaces.
How to install io.github.KuvopLLC/better-bear
Copy-paste configuration for popular MCP clients.
{
"mcpServers": {
"better-bear": {
"command": "npx",
"args": [
"-y",
"better-bear"
]
}
}
}Tools & capabilities
Tools this server exposes to the agent.
List notes— List all notes in your Bear librarySearch notes— Full-text search across notesCreate note— Create a new note with title and bodyEdit note— Edit or append text to an existing noteList tags— List all tags in your libraryAttach file— Attach a file to a noteLibrary statistics— Get statistics about your note libraryHealth check— Verify server and CloudKit connection statusContext library init— Set up a context library for curated notesContext library sync— Sync notes tagged with #contextContext library add— Add a note to the context library with optional subtagContext library status— Check context library health and status
Use cases
- Search your Bear notes and retrieve relevant information while working in Claude
- Create and edit notes directly from Claude without switching apps
- Build a curated context library of your most important notes for Claude to reference
- Organize and tag notes programmatically via AI
- Attach files and manage note metadata through Claude
io.github.KuvopLLC/better-bear MCP server FAQ
better-bear is an MCP server that connects Claude to your Bear notes via CloudKit. It lets you read, search, create, edit, and manage notes, plus set up a context library for efficient AI-powered note retrieval.
Yes, better-bear is open-source (MIT license) and free to use. It requires a Bear app subscription (separate) and iCloud account for CloudKit authentication.
Run `curl -sL https://raw.githubusercontent.com/KuvopLLC/better-bear/main/install.sh | bash`, then `bcli auth` and `bcli mcp install`. This sets up the MCP server for Claude Desktop and Claude Code.
You need an iCloud account (for CloudKit) and a Bear app subscription. The CLI guides you through iCloud authentication with `bcli auth`.
The context library is a synced folder of notes you tag with #context in Bear. Claude can efficiently query this curated subset without loading your entire library, following Karpathy's knowledge base pattern.
Yes, all MCP tools work as CLI commands too. For example: `bcli search "query"`, `bcli create "Title" -b "Body"`, `bcli tags`, and `bcli stats`.
README (reference)
Source of truth, from the repository.
better bear
MCP server and CLI for Bear notes via CloudKit. Includes a context library — a curated, synced folder of notes optimized for LLM consumption, inspired by Karpathy's LLM Knowledge Base pattern.
Full docs: better-bear.com
Install
Install the CLI, then connect to Claude:
curl -sL https://raw.githubusercontent.com/KuvopLLC/better-bear/main/install.sh | bash
bcli auth
bcli mcp install
This installs the bcli binary, authenticates with iCloud, and sets up the MCP server for both Claude Desktop (via .mcpb bundle) and Claude Code.
Other install methods
| Method | Command |
|---|---|
| Claude Desktop only | bcli mcp install --desktop-only |
| Claude Code only | bcli mcp install --code-only |
| Claude Code (direct) | claude mcp add better-bear -- npx -y better-bear |
| Config file | bcli mcp install --json |
| .mcpb bundle | Download from latest release and double-click |
Manage
bcli mcp status # check what's configured
bcli mcp uninstall # remove from Claude Desktop and Claude Code
bcli mcp reinstall # clean uninstall + install
bcli upgrade # upgrade bcli binary
CLI
All commands also work standalone from the terminal:
bcli ls # list notes
bcli search "query" # full-text search
bcli create "Title" -b "Body" # create a note
bcli edit <id> --append "text" # append to a note
bcli tags # list all tags
bcli attach <id> photo.jpg # attach a file
bcli stats # library statistics
bcli health # health check
See better-bear.com for the full command reference.
Context Library
Turn a subset of your Bear notes into a synced, curated context folder that Claude can navigate using index-first retrieval. Tag notes with #context in Bear, sync, and Claude reads a compact index to find relevant files — loading only what it needs, not everything.
bcli context init # one-time setup
bcli context sync # pull qualifying notes
bcli context add <id> --subtag research # tag a note for inclusion
bcli context status # health check
Or tell Claude: "Set up a context library" — and it handles everything via MCP tools.
The architecture follows Karpathy's three-folder pattern: bear/ (synced from CloudKit), external/ (PDFs, exports, shared docs), and inbox/ (drop zone for triage). An index.md manifest maps everything. See better-bear.com for full documentation.
MCP Tools
34 tools covering notes, tags, TODOs, attachments, search, front matter, stats, health checks, and the context library. See the MCP server README for the full list.
Contributors
<a href="https://github.com/mreider"><img src="https://avatars.githubusercontent.com/u/118036?v=4" width="50" height="50" style="border-radius:50%" alt="mreider"></a> <a href="https://github.com/program247365"><img src="https://avatars.githubusercontent.com/u/13910?v=4" width="50" height="50" style="border-radius:50%" alt="program247365"></a> <a href="https://github.com/asabirov"><img src="https://avatars.githubusercontent.com/u/733858?v=4" width="50" height="50" style="border-radius:50%" alt="asabirov"></a> <a href="https://github.com/darronz"><img src="https://avatars.githubusercontent.com/u/136805?v=4" width="50" height="50" style="border-radius:50%" alt="darronz"></a>
Related MCP servers

Kybase
Self-hosted Markdown memory for AI agents: notes you can read and edit, with hybrid search.

45+ interactive charts, dashboards, and KPIs rendered directly in AI conversations

io.github.KyuRish/patreon-mcp-server
Access your Patreon creator data - campaigns, patrons, tiers, and posts.

AI ops team for founders: approvals queue, inbox triage, briefing, follow-up drafting.
View repository →Fact-check any claim with live web sources. Ed25519-signed receipts. Catches hallucinations.

Epoch
Epoch time estimation MCP server for PERT, COCOMO, Monte Carlo, calibration, cost, and risk.
