hubmesh runs on your own machine — the client starts it, so there is no endpoint to ping. 83 installs a week from pypi. Last commit 19 Sep 2026.
Deterministic multi-hop graph retrieval for RAG. Zero LLM calls in the query path.
Today is the operative word: we check hubmesh every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
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 hubmesh -- uvx hubmesh
{
"mcpServers": {
"hubmesh": {
"args": [
"hubmesh"
],
"command": "uvx"
}
}
}
[mcp_servers.hubmesh]
command = "uvx"
args = ["hubmesh"]
{
"mcpServers": {
"hubmesh": {
"args": [
"hubmesh"
],
"command": "uvx"
}
}
}
{
"mcpServers": {
"hubmesh": {
"args": [
"hubmesh"
],
"command": "uvx"
}
}
}
Deterministic SVG perception for LLM agents — measures, never guesses. MCP server.
Local-first memory for AI agents: SQLite FTS5, deterministic recall, no vector DB, no cloud.
Single-binary local RAG MCP server in Rust. MiniLM compiled in, hybrid search, zero deps.
Local-first agentic RAG with citations - hybrid search, reranking, multimodal document retrieval
Local RAG MCP for markdown documentation. Retrieval only; host synthesizes answers.
Repository intelligence, code graph, history, papertrail, and cross-agent memory for coding agents.
Graph + vector memory for agents: recall, ingest, search, distill. Local or remote backend.
Private on-device memory & retrieval for AI assistants — offline vector + lexical search.
Answers built from our own checks of this server.