MCP server giving AI coding agents offline, token-efficient persistent memory — recall decisions, patterns, and bugs across sessions with 68.5% fewer tokens than re-reading context. No LLM, no server; works with 15+ agents.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add toon-memory --env TOON_MEMORY_KEY='<TOON_MEMORY_KEY>' --env GITHUB_TOKEN='<GITHUB_TOKEN>' -- npx -y toon-memory~/.codex/config.toml
[mcp_servers.toon-memory]
command = "npx"
args = ["-y", "toon-memory"]
env = { TOON_MEMORY_KEY = "<TOON_MEMORY_KEY>", GITHUB_TOKEN = "<GITHUB_TOKEN>" }
# codex mcp add toon-memory --env TOON_MEMORY_KEY='<TOON_MEMORY_KEY>' --env GITHUB_TOKEN='<GITHUB_TOKEN>' -- npx -y toon-memory~/.cursor/mcp.json
{
"mcpServers": {
"toon-memory": {
"command": "npx",
"args": [
"-y",
"toon-memory"
],
"env": {
"TOON_MEMORY_KEY": "<TOON_MEMORY_KEY>",
"GITHUB_TOKEN": "<GITHUB_TOKEN>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"toon-memory": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"toon-memory"
],
"env": {
"TOON_MEMORY_KEY": "<TOON_MEMORY_KEY>",
"GITHUB_TOKEN": "<GITHUB_TOKEN>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"toon-memory": {
"command": "npx",
"args": [
"-y",
"toon-memory"
],
"env": {
"TOON_MEMORY_KEY": "<TOON_MEMORY_KEY>",
"GITHUB_TOKEN": "<GITHUB_TOKEN>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"toon-memory": {
"command": "npx",
"args": [
"-y",
"toon-memory"
],
"env": {
"TOON_MEMORY_KEY": "<TOON_MEMORY_KEY>",
"GITHUB_TOKEN": "<GITHUB_TOKEN>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"toon-memory": {
"command": "npx",
"args": [
"-y",
"toon-memory"
],
"env": {
"TOON_MEMORY_KEY": "<TOON_MEMORY_KEY>",
"GITHUB_TOKEN": "<GITHUB_TOKEN>"
}
}
}
}Environment variables
TOON_MEMORY_KEYGITHUB_TOKEN
Replace the placeholders with your own credentials and keep them out of version control.