Skip to main content
Glama

📚 MCP Notes

A personal knowledge management system built on the Model Context Protocol (MCP) that transforms daily notes into organized, searchable knowledge.

🚀 Usage

🛠️ Manual Setup

  1. Clone the repository

    git clone git@github.com:markacianfrani/mcp-notes.git
    cd mcp-notes
    npm i
    npm run build
  2. Setup MCP

🖥️ Claude Desktop

Add to your claude-desktop-config.json:

"notes": {
    "command": "node",
    "args": [
        "path/to/this/repo/mcp-notes/dist/index.js",
        "path/to/your/notes"
    ],
},

Related MCP server: Memory Bank MCP

🌟 Project Vision

MCP Notes aims to solve the problem of knowledge fragmentation by creating a system where daily thoughts, accomplishments, and insights can be:

  1. 📥 Captured quickly with minimal friction

  2. 🗂️ Organized automatically into meaningful structures

  3. 🧠 Synthesized into higher-level knowledge

  4. 🔍 Retrieved easily when needed

The goal is to build an external rubber duck that can dump your working memory in real-time from any tool and sort through the garbage.

🧩 Core Concepts

1. 📅 Daily Logs

Daily logs are the atomic unit of capture in MCP Notes. Each day gets its own markdown file where you can record observations throughout the day. Think of daily logs like a running ledger.

💡 TIP: Make use of Claude Desktop's Quick Entry Keyboard Shortcut

MCP is tool-agnostic so you can /log across conversations but also tools as well--ask Copilot or Cursor to log a summary of lessons-learned after fixing a bug or make Claude Desktop save a new approach to an old problem.

2. 📊 Rollups

Rollups are automatically generated summaries that condense daily logs into higher-level insights and patterns. They help connect isolated pieces of information into a coherent narrative of your work and thinking.

3. 📚 Knowledge Categories

All notes in the system fall into one of four categories:

  • 🏆 Accomplishment: Solving problems, completing features, fixing bugs

  • 💡 Insight: Patterns, architectural decisions, better ways of working

  • 📝 TODO: Tasks connected to larger goals, meaningful improvements

  • 📖 Knowledge: Technical details, context, rationales, techniques

🎨 Design Principles

MCP Notes is built on several core design principles:

1. 📄 Plain Text First

All notes are stored as plain text Markdown files, ensuring:

  • Future-proof storage that won't be locked into proprietary formats

  • Version control compatibility

  • Easy editing with any text editor

  • Transparent data structure

2. ✍️ Low Friction Capture

The primary interface is designed to minimize the friction of recording thoughts:

  • Simple text input

  • Automatic categorization assistance

  • No complex organization required at capture time

3. 🔄 Progressive Organization

Rather than requiring rigid organization upfront, MCP Notes employs a progressive approach:

  • Capture raw thoughts and activities

  • Automatically categorize content

  • Generate periodic summaries

  • Connect related items over time

4. 🤖 AI Augmentation

The system leverages AI to enhance human thinking, not replace it:

  • Help categorize information

  • Generate summaries and connections

  • Surface relevant past notes

  • Identify patterns across time

🧰 Available Tools

MCP Notes provides a set of tools that can be invoked through Claude Desktop or other MCP-compatible clients. These tools allow you to capture, organize, and retrieve your notes.

📝 Core Note Tools

/log

Creates or updates today's daily log file with your notes.

Invoke with: "log this to my daily log: ...", "add a summary of that to my log: ...."

/rollup

Synthesizes daily notes into an organized summary with categories, connections, and action items.

Invoke with: "rollup my notes for today"

write_note

Creates a new note or overwrites an existing note with content.

sticky

Evaluates the "stickiness" of a thought based on actionability, longevity, findability, and future reference value.

Invoke with: "Is this idea sticky?"

evaluateInsight

Evaluates the long-term value and significance of an insight or thought.

Invoke with: "Is this insight valuable for the long term?"

📂 File System Tools

search_files

Recursively searches for files and directories matching a pattern in your notes directory.

read_note

Reads the complete contents of a note file from your notes directory.

read_multiple_notes

Reads the contents of multiple note files simultaneously.

list_directory

Lists the contents of a directory in your notes.

create_directory

Creates a new directory in your notes.

See the CHANGELOG.md file for version history and changes.

💡 Available Prompts

Is this atomic?

Breaks down ideas into their simplest standalone parts. Use this prompt to turn large ideas into smaller concepts. Smaller notes can be linked to other notes much more easily.

Available Tools

9 tools
create_directoryB

Create a new directory in your notes. Can create nested directories in one operation. Path should be relative to your notes directory.

Inputs: path (required).

evaluateInsightB
  Evaluate the long-term value and significance of an insight or thought based on the following criteria: 
  1. Actionability (1-10): Can this be applied to future work? Is there any information that can be used to apply this thought to future work in different contexts? Is the problem it solves clear?
  2. Longevity (1-10): Will this be relevant months or years from now?
  3. Findability (1-10): Would this be hard to rediscover if forgotten?
  4. Future Reference Value (1-10): How likely are you to need this again?
  
  Insights to ignore: 
  1. Trivial syntax details
  2. Redundant information
  

Inputs: actionability (required), evaluationStep (required), findability (required), futureReferenceValue (required), +4 more.

list_directoryA

List the contents of a directory in your notes. Shows all files and directories with clear labels. Specify path relative to your notes directory (e.g., 'Log' or 'Rollups').

Inputs: path.

logC

Create or update today's daily log file. Optionally add notes to the log.

Inputs: notes (required), tags (required).

read_multiple_notesA

Read the contents of multiple note files simultaneously. Specify paths relative to your notes directory (e.g., ['Log/2023-01-01.md', 'Rollups/2023-01-01-rollup.md']). Returns each file's content with its path as a reference.

Inputs: paths (required).

read_noteA

Read the complete contents of a note file from your notes directory. Specify the path relative to your notes directory (e.g., 'Log/2023-01-01.md'). Returns the full text content of the note file.

Inputs: path (required).

rollupA

Synthesize my daily note to create an organized rollup of the most important notes with clear categories, connections, and action items. Optionally specify a date (YYYY-MM-DD). Only include notes that actually add long-term value. If you are unsure, call the /evaluateInsight tool to evaluate the long-term value of the thought. If you do not have enough information, stop and ask the user for more information. It is better to not log anything than log something that is not useful.

Inputs: accomplishments (required), insights (required), date, todos.

search_filesA

Recursively search for files and directories matching a pattern in your notes directory. The search is case-insensitive and matches partial names. Returns full paths to all matching items. Great for finding notes when you don't know their exact location.

Inputs: pattern (required), excludePatterns.

write_noteC

Create a new note or overwrite an existing note with content. Path should be relative to your notes directory. Optionally include tags that will be merged with any existing tags in the note.

Inputs: content (required), path (required), tags.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.5/5.0

    Scored across 9 tools

    Disambiguation4/5

    Most tools have distinct purposes, such as create_directory for directory creation, read_note for reading single files, and rollup for synthesizing daily notes. However, there is some overlap between read_note and read_multiple_notes, as both handle reading files, which could cause minor confusion, but their descriptions clarify the distinction (single vs. multiple files).

    Naming Consistency3/5

    The naming conventions are mixed, with most tools using snake_case (e.g., create_directory, list_directory) and one using camelCase (evaluateInsight). This inconsistency reduces predictability, but the names are still readable and generally follow a verb_noun pattern, except for evaluateInsight which deviates in style.

    Tool Count5/5

    With 9 tools, the count is well-scoped for a notes management server, covering core operations like creating, reading, writing, listing, searching, and synthesizing notes. Each tool serves a clear purpose, and there is no bloat or missing essential functions, making the set appropriately sized for the domain.

    Completeness4/5

    The tool set provides comprehensive coverage for notes management, including CRUD-like operations (create, read, write), directory handling, searching, and advanced features like rollup and insight evaluation. A minor gap exists in update operations for notes or directories, as write_note only allows overwriting, but agents can work around this with existing tools.

    Maintenance

    ActivityInactive
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      D
      maintenance
      A personal knowledge management system built on the Model Context Protocol that transforms daily notes into organized, searchable knowledge.
      9
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      A Model Context Protocol plugin that helps AI assistants maintain persistent project context through structured markdown files, providing a systematic approach to tracking project goals, decisions, progress, and patterns.
      3
      18 npm
      15
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      A personal knowledge management MCP server that enables capturing daily notes, meeting summaries, and ideas through Claude, with full-text search, git sync, and git activity tracking across projects.
      10 npm
      2
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      An AI-optimized knowledge management system that provides semantic search, memory storage, and intelligent deduplication through the Model Context Protocol (MCP).
      68 npm
      2
      MIT