Local-first memory layer for LLM apps and agents. Runs offline on Ollama, sentence-transformers and FAISS; OpenAI and Anthropic optional. Importance scoring, contradiction resolution, confidence-aware retrieval.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add widemem-ai -- python -m widemem.mcp_server~/.codex/config.toml
[mcp_servers.widemem-ai]
command = "python"
args = ["-m", "widemem.mcp_server"]
# codex mcp add widemem-ai -- python -m widemem.mcp_server~/.cursor/mcp.json
{
"mcpServers": {
"widemem-ai": {
"command": "python",
"args": [
"-m",
"widemem.mcp_server"
]
}
}
}.vscode/mcp.json
{
"servers": {
"widemem-ai": {
"type": "stdio",
"command": "python",
"args": [
"-m",
"widemem.mcp_server"
]
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"widemem-ai": {
"command": "python",
"args": [
"-m",
"widemem.mcp_server"
]
}
}
}claude_desktop_config.json
{
"mcpServers": {
"widemem-ai": {
"command": "python",
"args": [
"-m",
"widemem.mcp_server"
]
}
}
}.trae/mcp.json
{
"mcpServers": {
"widemem-ai": {
"command": "python",
"args": [
"-m",
"widemem.mcp_server"
]
}
}
}