Persistent layered memory MCP server for AI agents — SQLite + FTS5 + vector hybrid search (RRF), multilingual, zero-cloud
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add mnemon --env MNEMON_EMBEDDING_API_KEY='<MNEMON_EMBEDDING_API_KEY>' -- npx -y mnemon-mcp~/.codex/config.toml
[mcp_servers.mnemon]
command = "npx"
args = ["-y", "mnemon-mcp"]
env = { MNEMON_EMBEDDING_API_KEY = "<MNEMON_EMBEDDING_API_KEY>" }
# codex mcp add mnemon --env MNEMON_EMBEDDING_API_KEY='<MNEMON_EMBEDDING_API_KEY>' -- npx -y mnemon-mcp~/.cursor/mcp.json
{
"mcpServers": {
"mnemon": {
"command": "npx",
"args": [
"-y",
"mnemon-mcp"
],
"env": {
"MNEMON_EMBEDDING_API_KEY": "<MNEMON_EMBEDDING_API_KEY>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"mnemon": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"mnemon-mcp"
],
"env": {
"MNEMON_EMBEDDING_API_KEY": "<MNEMON_EMBEDDING_API_KEY>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"mnemon": {
"command": "npx",
"args": [
"-y",
"mnemon-mcp"
],
"env": {
"MNEMON_EMBEDDING_API_KEY": "<MNEMON_EMBEDDING_API_KEY>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"mnemon": {
"command": "npx",
"args": [
"-y",
"mnemon-mcp"
],
"env": {
"MNEMON_EMBEDDING_API_KEY": "<MNEMON_EMBEDDING_API_KEY>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"mnemon": {
"command": "npx",
"args": [
"-y",
"mnemon-mcp"
],
"env": {
"MNEMON_EMBEDDING_API_KEY": "<MNEMON_EMBEDDING_API_KEY>"
}
}
}
}Environment variables
MNEMON_EMBEDDING_API_KEY
Replace the placeholders with your own credentials and keep them out of version control.