Documentation

RoboTunnel Docs

RoboTunnel is built around one narrow promise: ROS 2 remote debugging that actually works. Start with the managed debug-session workflow, then go deeper into Debug Projection, CLI commands, Discord workflows, trust boundaries, and architecture.

Primary path CLI-first onboarding and managed debug sessions.

Differentiator Debug Projection for real-world field links.

Trust model Open-source agent, local keys, explicit auth.

Getting Started

The shortest path from developer machine + robot to your first managed debug session.

CLI install, agent install, first session

Debug Projection

The core differentiated capability: profiles, topic policy, supported topic classes, and tool integration.

Foxglove, RViz, stream pull, policy overrides

Skills

The four first-touch robot-side skills: visual_debug, ros2_observe, host_debug, and monitor.

Capabilities, examples, and when to use each path

CLI Reference

Reference for init, list, debug, connect, and robot-side skill invocation.

Primary developer command surface

Discord Bot

How Discord fits into RoboTunnel, what it can do, and what intentionally stays CLI-first.

Operators, alerts, natural language

Troubleshooting

What to check when WebRTC, signaling, or debug attach does not come up cleanly.

Quick checks, preflight, and next-step diagnosis

Security

Trust boundary, local LLM key handling, Ed25519 auth, and the open-source edge model.

Security and data-handling stance

Architecture

How CLI, Discord, platform, and robot-side agent fit together across the four-layer model.

Connection, platform, application, interaction