RivetOS: run your own agents, own the memory they share

01 Runs on your hardware

You install it. Docker on a laptop, LXC on Proxmox, or the local one-liner on Linux and macOS. Providers are plugins: Anthropic, xAI, Google, or fully local via Ollama, vLLM, and llama.cpp. Your keys, your GPUs, your machines.

anthropicxaigoogleollamavllmllama.cppclaude-clicodex-climcpagent-http

02 Local-first memory

The memory plane lives on the hub you run: PGlite on one machine, or Postgres on a datahub you enroll nodes into. Hybrid full-text + vector search, summary compaction, cross-agent recall over MCP. The record is yours.

rivet> memory_search "mesh CA decision?"
● 3 hits · hybrid fts+vector · 48ms
  1. 06-12 one root CA, per-node certs
  2. 06-13 rotation via update --mesh
  3. 06-27 roster fix deployed fleet-wide

03 A mesh you enroll

One machine is enough. Add nodes when you want them: different models, one shared memory, mTLS between machines you control. Same runtime on every node. There is no Rivet-operated mesh to join.

Start on your machine. Read the rest here.

$ curl -fsSL https://get.rivethub.io/local.sh | bash

RivetHub is the get-started surface (agent brief + local installer). From a source checkout, runnpx rivetos local. Docker, mesh, and a dedicated datahub are day-2.Getting started →