yggdrasil — Claude Code MCP server: memory and install
A local memory layer for AI coding agents that carries project decisions and status across sessions and tools.
claude mcp add --env YGG_ENGINE_URL=<YGG_ENGINE_URL> --env YGG_ENGINE_TOKEN=<YGG_ENGINE_TOKEN> --transport stdio yggdrasil -- uvx yggdrasil-memoryWhat yggdrasil does
yggdrasil gives Claude Code, Codex, Cursor, and other MCP hosts one shared local memory. An agent can store a project decision, recall an earlier fix, or retrieve a related note from another project in a later session. The data stays on the local machine.
The server runs through uvx and reads YGG_ENGINE_URL plus YGG_ENGINE_TOKEN. Claude Code adds it under the name yggdrasil; Cursor and GitHub Copilot use JSON configuration. The README describes a locally generated token and says an API key is not required. The installation command still expects values for the token and engine URL variables.
What it helps with
- carry project decisions into a new session
- recall project status during a fresh chat
- retrieve a related fix from another project
- share memory across MCP hosts
Example prompts for the agent
- Remember the decision we made about the database migration for future sessions.
- Recall how we fixed a similar error in another project.
- What is the current status we saved for this project?
How to install yggdrasil
Platform CLI
claude mcp add --env YGG_ENGINE_URL=<YGG_ENGINE_URL> --env YGG_ENGINE_TOKEN=<YGG_ENGINE_TOKEN> --transport stdio yggdrasil -- uvx yggdrasil-memoryReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from yggdrasil-memory <command>
Platform CLI
codex mcp add yggdrasil --env YGG_ENGINE_URL=<YGG_ENGINE_URL> --env YGG_ENGINE_TOKEN=<YGG_ENGINE_TOKEN> -- uvx yggdrasil-memoryReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from yggdrasil-memory <command>
Config file
{
"mcpServers": {
"yggdrasil": {
"command": "uvx",
"args": [
"yggdrasil-memory"
],
"env": {
"YGG_ENGINE_URL": "<YGG_ENGINE_URL>",
"YGG_ENGINE_TOKEN": "<YGG_ENGINE_TOKEN>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from yggdrasil-memory <command>
Config file
{
"servers": {
"yggdrasil": {
"type": "stdio",
"command": "uvx",
"args": [
"yggdrasil-memory"
],
"env": {
"YGG_ENGINE_URL": "<YGG_ENGINE_URL>",
"YGG_ENGINE_TOKEN": "<YGG_ENGINE_TOKEN>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from yggdrasil-memory <command>
Platform CLI
gemini mcp add -e YGG_ENGINE_URL=<YGG_ENGINE_URL> -e YGG_ENGINE_TOKEN=<YGG_ENGINE_TOKEN> yggdrasil uvx yggdrasil-memoryReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from yggdrasil-memory <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about yggdrasil
How do I install yggdrasil in Claude Code?
Run claude mcp add yggdrasil -e YGG_ENGINE_TOKEN=<YGG_ENGINE_TOKEN> -- uvx yggdrasil-memory.
What does yggdrasil do?
It connects AI coding agents to one local memory. Project decisions, lessons, and status can remain available across sessions, tools, and projects.
Does yggdrasil require an API key?
No API key is required. The engine creates a local token for its memory service.
