Skip to content
MCP server

Enzyme

Local concept search over Obsidian and markdown notes for your coding agent

Runs locallyStatus and history
Pricing
Freemium
Platforms
macOS, Linux, CLI
Runs
Locally
Version
0.13.4
Listing updated
3 days ago

About Enzyme

Enzyme is a freemium MCP server for knowledge and memory. It runs locally on your machine and is started by your AI client.

Enzyme builds a local index over a folder of markdown notes and lets an agent search it by theme rather than by exact words. During setup it samples your notes across time, using tags, wikilinks and folders as signals, then generates background questions it calls catalysts. Each catalyst is embedded, so a query lands on the notes nearest to the questions they answer.

The agent works through a small command set: enzyme petri returns current entities and catalysts as working memory, and enzyme catalyze "query" searches by concept. An .enzyme program, proposed from your notes and agent instructions, controls which folders and tags to learn from; you review it before the index is built. Your notes stay where they are and are not modified.

The README says the approach saves up to 50% of tokens compared with plain agentic grep search. It targets fast-growing collections such as Obsidian vaults, meeting notes, Markdown CRM setups and agent memory written as Markdown. The registry entry ships macOS (arm64, x86_64) and Linux (x86_64, arm64) bundles that read the vault path from ENZYME_VAULT_PATH.

Install

Run in a terminal

claude mcp add enzyme -- npx -y https://github.com/byenzyme/enzyme/releases/download/v0.13.4/enzyme-mcp-0.13.4-macos-arm64.mcpb

Add to claude_desktop_config.json

{
  "mcpServers": {
    "enzyme": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/byenzyme/enzyme/releases/download/v0.13.4/enzyme-mcp-0.13.4-macos-arm64.mcpb"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "enzyme": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/byenzyme/enzyme/releases/download/v0.13.4/enzyme-mcp-0.13.4-macos-arm64.mcpb"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "enzyme": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/byenzyme/enzyme/releases/download/v0.13.4/enzyme-mcp-0.13.4-macos-arm64.mcpb"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "enzyme": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/byenzyme/enzyme/releases/download/v0.13.4/enzyme-mcp-0.13.4-macos-arm64.mcpb"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "enzyme": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/byenzyme/enzyme/releases/download/v0.13.4/enzyme-mcp-0.13.4-macos-arm64.mcpb"
      ]
    }
  }
}

Configuration

  • ENZYME_VAULT_PATHPath to your Obsidian/Markdown vault directory
  • ENZYME_VAULT_PATHPath to your Obsidian/Markdown vault directory
  • ENZYME_VAULT_PATHPath to your Obsidian/Markdown vault directory
  • ENZYME_VAULT_PATHPath to your Obsidian/Markdown vault directory

What it does

Catalyst-based retrieval
Background-generated questions are embedded so a query finds notes by theme, not only by keywords.
Working memory command
`enzyme petri` returns current entities and catalysts compiled from the note trail.
Concept search
`enzyme catalyze` searches the vault by concept using catalyst vocabulary.
Reviewable setup program
A proposed `.enzyme` file lists which folders, tags and periods to learn from, and you can correct it first.
Incremental refresh
New notes are ingested locally and quickly, while catalysts evolve on a slower, cheaper cycle.
Notes left untouched
The index is built locally and existing notes stay in place and unchanged.

Good for

  • Give a coding agent context on decisions spread across an Obsidian vault
  • Find connections between meeting notes and interview write-ups that never reference each other
  • Keep agent memory written as markdown searchable across sessions
  • Cut token use compared with grepping a large notes folder
  • Surface themes from dated notes in a Markdown CRM

Pricing

Freemium, from $19 a month

Free plan with 30 credits; Pro listed at $19/mo; custom plans for products

Prices change. Check memory.enzyme.garden before you buy.

Enzyme FAQ

Do I need an account?

Yes for the current CLI flow. Login provides the auth used for hosted credits and the workspace dashboard; vault files stay on your machine unless you publish a hosted vault.

What does it cost?

A free plan includes 30 credits and unlimited retrieval, and Pro is listed at $19 per month with 200 credits and hosted MCPs. Custom pricing covers product deployments.

Does my data leave my machine?

The index runs locally. Catalyst generation sends selected excerpts through Enzyme hosted credits by default, or to your own LLM provider when you pass --use-env-llm.

How are credits used?

New accounts start with 30 credits. By default `enzyme init` uses 3 credits and `enzyme refresh` uses 0.5.

Is Enzyme free?

Enzyme has a free plan, with paid plans for more usage or features. Paid plans start at $19 a month.

How do I add Enzyme to Claude Code?

Run this command in your terminal: claude mcp add enzyme -- npx -y https://github.com/byenzyme/enzyme/releases/download/v0.13.4/enzyme-mcp-0.13.4-macos-arm64.mcpb

What are the best alternatives to Enzyme?

The closest alternatives to Enzyme are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and note1.

Alternatives to Enzyme

  1. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  2. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  3. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  4. Local-first long-term memory layer for AI agents, served over MCP

    Not checked yet
    • 5 tools
    • Remote
    • Open source
    • 1.3K GitHub stars
    • Knowledge and memory
  5. note1

    MCP server

    MCP server giving AI assistants search and summaries across your note1 meetings

    • 4 tools
    • Remote
    • Free trial
    • 1 GitHub star
    • Productivity