Yggdrasil

by vondervuflya

288 downloads
Not rated
GitHub

About

Durable, local-first memory for AI coding agents over MCP — zero-dependency, curated & semantically de-duped, you own the data (SQLite + Markdown). Works with Claude Code, Codex & any MCP host.

Details

Author
vondervuflya
Downloads
288
Categories
Other, Developer Tools, AI, Knowledge Base

- Persistent memory survives across AI sessions and projects
- One brain shared by Claude Code, Codex, and any MCP host
- Cross-project recall (“this looks like project B — reuse it?”)
- Curated, not captured: agent saves only what matters; governance dedupes and archives
- Opt-in self-maintenance using a small local model (zero API tokens)
- 100% local, no cloud, no account, no telemetry

Setting up with Highlight

This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:

  1. Download and install Highlight from highlightai.com/download
  2. Navigate to the plugins tab and select "Add Custom Plugin"
  3. Configure the plugin with the settings below
    Plugin Name Yggdrasil
    Command (node, npx, python, etc.)

    Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.

  4. Enable "Start Automatically" if you want the plugin to start when Highlight launches

From the repository

Install inside Claude Code with two commands (/plugin marketplace add VonderVuflya/Yggdrasil then /plugin install yggdrasil) or via CLI using uvx --from yggdrasil-memory ygg install, npx yggdrasil-memory install, pipx, pip, Homebrew (brew install VonderVuflya/tap/yggdrasil), or a .mcpb file for Claude Desktop. Once running, agents interact with it through MCP tools: ygg_health, ygg_bootstrap, ygg_search, ygg_recall, ygg_remember, and ygg_materialize. Hooks automatically inject a project status summary (~300 tokens) at the start of each session.

Claude Desktop / Cursor

Paste into your MCP client config file to install this server.

{
    "mcpServers": {
        "yggdrasil": {
            "yggdrasil": {
                "command": "uvx",
                "args": [
                    "--from",
                    "yggdrasil-memory",
                    "ygg",
                    "mcp"
                ],
                "env": {
                    "YGG_ENGINE_URL": "http://127.0.0.1:42069",
                    "YGG_ENGINE_TOKEN": "YOUR_TOKEN"
                }
            }
        }
    }
}

McpServers

{
    "yggdrasil": {
        "command": "uvx",
        "args": [
            "--from",
            "yggdrasil-memory",
            "ygg",
            "mcp"
        ],
        "env": {
            "YGG_ENGINE_URL": "http://127.0.0.1:42069",
            "YGG_ENGINE_TOKEN": "YOUR_TOKEN"
        }
    }
}

🌳 Yggdrasil

Stop re-explaining your project to every new AI session. One local memory for Claude Code, Codex, and every MCP agent — shared across sessions, tools, and projects. Zero dependencies. Nothing leaves your machine. Every new chat, your AI forgets. You re-explain the project, the decisions, the gotchas — every time, in every tool. Yggdrasil is a tiny always-on memory any agent plugs into: open a new session, in any project, with any AI, and it already knows what you decided, what broke, and what's still open.

Install

Inside Claude Code (two commands): /plugin marketplace add VonderVuflya/Yggdrasil /plugin install yggdrasil Or via CLI (registers MCP tools with Claude Code and Codex): uvx --from yggdrasil-memory ygg install Also available via npx yggdrasil-memory install, pipx, pip, Homebrew (brew install VonderVuflya/tap/yggdrasil), and a .mcpb for Claude Desktop.

Why

- 🧠 Persistent — decisions, lessons, and project status survive across sessions - 🔌 One brain, every tool — Claude Code, Codex, and any MCP host share the same memory - 🌐 Cross-project recall"this looks like what you did in project B — reuse it?" - 🧹 Curated, not captured — the agent saves the few things that matter; governance dedupes and archives, never deletes - 🌱 Self-maintaining (opt-in) — a small local model consolidates memory in the background, zero API tokens - 🔒 100% local — no cloud, no account, no telemetry

How it works

A tiny always-on daemon (~21 MB RAM) your agents reach over MCP tools: ygg_health, ygg_bootstrap, ygg_search, ygg_recall, ygg_remember, ygg_materialize. Hooks auto-inject project status at session start (~300 tokens). Runs on SQLite + FTS5 with zero dependencies out of the box; optional local embeddings via Ollama lift recall@1 from 0.77 to 0.93 (recall@3 = 1.00, reproducible benchmark included). Every memory doubles as a plain-Markdown note you can read in Obsidian. Source & docs: https://github.com/VonderVuflya/yggdrasil · License: AGPL-3.0
No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.