Memharness runs on your own machine — the client starts it, so there is no endpoint to ping. 45 installs a week from npm. Last commit 26 Jun 2026.
Bi-temporal, provenance-carrying long-term memory for AI agents, in one SQLite file.
Nothing serious here today
Today is the operative word: we check Memharness every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
Connect this server
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add memharness -- npx -y @memharness/mcp{
"mcpServers": {
"memharness": {
"command": "npx",
"args": [
"-y",
"@memharness/mcp"
]
}
}
}[mcp_servers.memharness]
command = "npx"
args = ["-y", "@memharness/mcp"]{
"mcpServers": {
"memharness": {
"command": "npx",
"args": [
"-y",
"@memharness/mcp"
]
}
}
}{
"servers": {
"memharness": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@memharness/mcp"
]
}
}
}Optional settings from the registry entry, left out of the block above:
MEMHARNESS_DB (Path to the SQLite memory file. Defaults to a per-user data path; set to override).
Alternatives to Memharness
same job, measured the same wayLocal-first memory for AI coding agents. Memory + CodeGraph + Wiki in one SQLite file.
Local long-term memory MCP server (SQLite) for AI coding agents — OpenCode, Claude Code, Cursor
Long-term memory and context fitting for agents, with no LLM in the loop. Local SQLite.
Persistent memory for AI coding agents: one Rust binary, one local SQLite database.
Local-first memory for AI agents: on-device hybrid retrieval over a single SQLite file.
Local-first memory for coding agents — MCP server, single SQLite file, local embeddings
Lightweight persistent memory for AI agents. Hybrid search. 16 tools. One SQLite file.
Persistent local memory for MCP-compatible AI coding agents, stored per repository in SQLite.
Memharness — questions
Answers built from our own checks of this server.