MCP Catalogs
Homeflywheel-memory screenshot

flywheel-memory

by velvetmonkey·44·Score 45

Local-first MCP server that provides AI agents with a knowledge graph over Obsidian vaults.

knowledge-graphproductivityai-llm
8
Forks
0
Open issues
this month
Last commit
2d ago
Indexed

Overview

Flywheel turns your Obsidian vault into a local memory layer for AI agents. It provides grounded search across notes with context, safe bounded writes back into the vault, and local indexes that stay on your machine. The system uses a 13-layer scoring mechanism that learns and improves over time. All processing happens locally with zero cloud dependencies, keeping your data private while giving AI agents access to structured knowledge about your notes.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:AI assistants searching across personal knowledge bases with contextual understanding
you:Automated note management with wikilink suggestions and entity linking
you:Policy-based vault automation for recurring tasks and follow-ups
you:How does Flywheel handle privacy?
you:Can I use Flywheel with other clients besides Claude?

When to choose this

Choose Flywheel when you need an AI memory layer for Obsidian or plain Markdown files, with local-first architecture, safe bounded writes, and optional semantic search.

When NOT to choose this

Don't choose Flywheel if you need cloud-based memory access, require write access to non-Markdown formats, or need multi-user collaborative features.

Tools this server exposes

5 tools extracted from the README
  • vault_search

    Search the vault with keywords or filters to find relevant notes

  • edit_section

    Edit or add content to a specific section in a note

  • policy

    Create, preview, or execute vault policies that search and modify notes

  • memory

    Access or update the memory system, including briefings and observations

  • graph

    Export or analyze the knowledge graph of notes and links

Note: Tool names extracted from descriptions and examples in the README, but without detailed signatures. The README mentions multiple tools but doesn't provide a complete list or explicit documentation of all MCP tools.

Comparable tools

memgraph-mcpsemantic-search-mcpobsidian-api-mcp

Installation

Quick Install

  1. Install as an Agent Skill:

``bash npx -y skills add velvetmonkey/flywheel-memory -g ``

  1. Wire the MCP server:

``bash bash <(curl -fsSL https://raw.githubusercontent.com/velvetmonkey/flywheel-memory/main/skills/flywheel/scripts/install.sh) ``

  1. Restart your client from your vault directory.

Manual MCP Configuration

Add to your .mcp.json:

{
  "mcpServers": {
    "flywheel": {
      "command": "npx",
      "args": ["-y", "@velvetmonkey/flywheel-memory"]
    }
  }
}

FAQ

How does Flywheel handle privacy?
All processing happens locally with zero cloud dependencies. Your notes remain readable Markdown, and Flywheel's state is stored locally beside your vault.
Can I use Flywheel with other clients besides Claude?
Yes, Flywheel works with any MCP-compatible client. Manual configuration is required for non-Claude clients like Cursor, Windsurf, and VS Code.

Compare flywheel-memory with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.