Shared, persistent memory for AI agents. Self-hosted MCP server with semantic search, vector RAG, and live updates. Works with Claude, Cursor, Codex, and any MCP client.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add memocat-shared-ai-memory --env MONTYCAT_URI='<MONTYCAT_URI>' -- uvx memocat-mcp~/.codex/config.toml
[mcp_servers.memocat-shared-ai-memory]
command = "uvx"
args = ["memocat-mcp"]
env = { MONTYCAT_URI = "<MONTYCAT_URI>" }
# codex mcp add memocat-shared-ai-memory --env MONTYCAT_URI='<MONTYCAT_URI>' -- uvx memocat-mcp~/.cursor/mcp.json
{
"mcpServers": {
"memocat-shared-ai-memory": {
"command": "uvx",
"args": [
"memocat-mcp"
],
"env": {
"MONTYCAT_URI": "<MONTYCAT_URI>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"memocat-shared-ai-memory": {
"type": "stdio",
"command": "uvx",
"args": [
"memocat-mcp"
],
"env": {
"MONTYCAT_URI": "<MONTYCAT_URI>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"memocat-shared-ai-memory": {
"command": "uvx",
"args": [
"memocat-mcp"
],
"env": {
"MONTYCAT_URI": "<MONTYCAT_URI>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"memocat-shared-ai-memory": {
"command": "uvx",
"args": [
"memocat-mcp"
],
"env": {
"MONTYCAT_URI": "<MONTYCAT_URI>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"memocat-shared-ai-memory": {
"command": "uvx",
"args": [
"memocat-mcp"
],
"env": {
"MONTYCAT_URI": "<MONTYCAT_URI>"
}
}
}
}Environment variables
MONTYCAT_URI
Replace the placeholders with your own credentials and keep them out of version control.