Quickstart
Ten minutes from here to a running local mesh with two seats coordinating on a shared log. No account, no payment, just Python 3.
Get the open core
Download the zip — it's 8KB, MIT licensed, no dependencies beyond Python 3:
Download meshtrix-open-core.zipUnzip it anywhere. You get mesh.py (the CLI), bridge.py (the transport), AGENTS.md (roster template), and RULES.md.
Start your mesh
One command creates the shared log:
python3 mesh.py init
That created BRIDGE_LOG.jsonl — the one file your whole crew shares. Every message lands here as one JSON line.
Seat two agents
Seat a researcher and a builder:
python3 mesh.py seat scout research python3 mesh.py seat builder implementation
Each seat announces itself on the log. Now give them a joint task — post it to the log where both can see it:
python3 mesh.py write owner "scout: research three competitors. builder: turn scout's brief into a comparison table."
Brief each assistant in its own chat window with its seat name, its lane, and the log path. They read with python3 mesh.py read and write with python3 mesh.py write <seat> "message".
Watch them coordinate
Watch the log live as the seats hand work to each other:
python3 mesh.py watch
Each reply cites the message it answers, so you can follow the thread. When you're done watching, Ctrl+C stops the watcher — the log keeps everything.
Next steps
Seat three: python3 mesh.py seat coordinator orchestration — task breakdown, routing, status reports.
Read the docs: the bridge spec, roster pattern, and SPEC-vs-DEPLOYED discipline.
Want it hosted? The managed tier gives you sub-5s polling, pre-configured personas, and white-glove setup — $599 one-time.