Lethe runs on your own machine — the client starts it, so there is no endpoint to ping. 110 installs a week from pypi. Last commit 10 Jul 2026.
Context GC for LLM agents: offload large tool outputs and recall them to save tokens.
Nothing serious here today
Today is the operative word: we check Lethe 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 lethe -- uvx lethe-llm-context{
"mcpServers": {
"lethe": {
"args": [
"lethe-llm-context"
],
"command": "uvx"
}
}
}[mcp_servers.lethe]
command = "uvx"
args = ["lethe-llm-context"]{
"mcpServers": {
"lethe": {
"args": [
"lethe-llm-context"
],
"command": "uvx"
}
}
}{
"mcpServers": {
"lethe": {
"args": [
"lethe-llm-context"
],
"command": "uvx"
}
}
}This one needs environment variables set before it will start:
LETHE_DB (Path to the SQLite archive file (default ./lethe.db)).
The author declared them in the registry entry; get the values from the project itself.
Alternatives to Lethe
same job, measured the same wayRun efficiency for coding agents: share of paid tokens that became output, not context re-reading.
Shared long-term memory for AI agents: save and recall context as a searchable knowledge graph.
JSON contract registry and validator for structured LLM and agent outputs.
Universal memory for AI agents and tools. Save, organize and search context anywhere.
Token and context budget for AI agents. Tracks spend and gates the session.
Local personal memory for agents as MCP tools — remember, recall, search, context, graph traversal.
Persistent memory for AI agents — log and recall conversation context over MCP.
Context selection and compression for AI agents - recall measured, not claimed.
Lethe — questions
Answers built from our own checks of this server.