A future-proof network stack for private systems

rstream is a secure tunnels platform for local services, private infrastructure, remote machines, devices, and internal systems. It creates controlled network paths without requiring inbound firewall rules, public IP exposure, or fragile access setups.
The founding idea behind rstream is simple. Modern applications have moved beyond a purely HTTP-shaped network. They carry web requests, realtime streams, browser sessions, terminal access, device control, background work, AI agents, private services, custom protocols, and long-lived machine-to-machine traffic. These paths differ in latency, state, reliability, and security requirements, yet they need a shared foundation for security, authentication, policy, observability, and automation.
Adding transport after transport without a common operating layer fragments identity, weakens policy consistency, makes logs harder to compare, and forces teams to rebuild the same access model for each new path. rstream brings these paths into one network layer. Agents connect outbound to the runtime, published endpoints and private tunnels route traffic back to upstream systems, and access is evaluated before it reaches the private side.
A wide protocol palette
rstream is built for the protocol surface used by modern systems. HTTP/1.1, HTTP/2, HTTP/3, WebSocket, WebTransport, TCP, TLS, QUIC, DTLS, STUN, TURN, WebRTC, SCTP, bytestreams, datagrams, WebTTY, and SDK integrations fit into the same connectivity model. The stack preserves transport-specific behavior where it affects latency, reliability, routing, browser semantics, or stream lifetime, while keeping identity, policy, logs, and automation consistent.
A short API request and a long-lived terminal stream should not require different security assumptions. A realtime media path, a device control channel, and a custom service should not create separate operational workflows. rstream gives these paths a common access model while preserving the protocol behavior that makes each workload useful in production.
Security and performance
rstream keeps upstream systems private by default. The private side initiates outbound sessions, and the runtime applies access controls before traffic reaches the upstream service. On the access side, rstream supports scoped tokens, rstream Auth, mutual TLS, challenge flows, IP and geographic policy, workspace protection, and trusted devices. On the cryptographic side, the direction includes end-to-end encryption beyond terminal workflows, post-quantum cryptography, platform keychains, security enclaves, and external HSMs.
The code that runs on customer devices is open source by policy. Agents, SDKs, client-side components, and local tooling can be inspected, embedded, and audited because users should be able to review the software that executes inside their own environments.
Performance receives the same level of attention. Secure access has to remain fast for interactive systems, realtime workloads, and long-running connections. rstream is built around persistent agent sessions, low-latency routing, protocol-aware dispatch, explicit error paths, connection logs, runtime metrics, and predictable reconnection behavior. The network path is designed to stay inspectable across admission, routing, shutdown, and audit.
Designed for Real Systems
rstream can begin with one CLI command that exposes a local service, then grow into private tunnels, persistent remote access, device fleets, customer environments, and deeply integrated platform deployments. A first setup can be driven from the CLI. A mature deployment can use SDKs, Docker labels or configuration, YAML runtime files, and Kubernetes operators, depending on how the system is already deployed.
That range lets existing stacks adopt rstream gradually. A developer can expose a local service quickly, an operator can standardize access to private machines, an integrator can deploy customer environments, and an enterprise team can align security and observability across more complex infrastructure. The same foundation carries the product from quick setup to production workloads without changing the core model of controlled connectivity.
About the Founder
Founded by @uartnet
software engineer
rstream is founded by @uartnet, a software engineer focused on networking, embedded systems, distributed infrastructure, and automation. The project reflects a long-standing interest in network protocols, remote systems, and infrastructure tooling, shaped by years of working with environments where access, transport, identity, and observability were handled by separate tools despite belonging to the same operational problem.
rstream is developed with feedback from technical users, operators, customers, security reviewers, and advisors. Their role is to challenge the product against real deployments and keep the work grounded in practical constraints. The ambition is a network stack that remains secure, fast, inspectable, and understandable from the first local tunnel to complex infrastructure deployments.