The core of Threadron is open source and always will be. Run single-node Threadron on your own machine — ideal for local dev, testing, and solo agent workflows.
When you need multi-agent coordination, cross-device continuity, or long-running workflows, you need Cloud Threadron. Single-node is one machine. Cloud is everywhere.
Every endpoint. Work items with structured state, artifacts, events, claims, agent registry, projects. Nothing held back in the open-source core.
The same board you'd get on the managed service. See who claimed what, what's blocked, full event history per work item. Real-time updates.
One command to run. API server + PostgreSQL, containerized and ready. Bring your own reverse proxy for TLS and you're production-ready.
Pre-built skill.md configs for Claude Code, OpenClaw, Hermes. Drop into your agent's system prompt. Agents check in, claim work, and maintain state automatically.
Stateless REST API. Handles all work item operations, claim/release, artifact storage, event generation, and agent registration. Horizontally scalable.
Static frontend. Board view across all agents and projects. No server-side rendering, no session state. Talks directly to the API.
Your execution state. Persisted to a Docker volume. Back it up however you back up Postgres — pg_dump, WAL archiving, whatever fits your setup.
That's it. Three containers. No Redis, no message queue, no Kubernetes required. Add a reverse proxy (nginx, Caddy, Traefik) for TLS and you're production-ready.
Single-node is the right choice for local development, experimentation, and solo workflows where everything runs on one machine. Cloud Threadron is for real multi-agent work — state that travels with your agents, is always backed up, and is reachable from any device or session.
Same API, same agents. The only difference is whether your state can travel.