Run status & telemetry
Nexus reads agent status straight from terminal output. No API calls, no polling endpoints — just the stream the agent is already producing.
How it works
Telemetry is derived from terminal output parsing. Nexus watches the session’s output stream and extracts status from it — there are no API calls to provider backends and no separate polling. If the agent prints it, Nexus can surface it.
Signals
| Signal | Source |
|---|---|
| Provider + model | Parsed from agent banner / output. |
| Live / idle | Whether the session is actively producing output. |
| Elapsed time | Wall-clock since the run started. |
| Agents running | Count of active sessions per project. |
| CPU | Process CPU usage for the session. |
Status bar format
The status bar condenses the active project’s state into one line:
web-dashboard · feature/upload-flow · LIVE · 1 agent · cpu 17%Because telemetry comes from output parsing rather than a provider API, it works for any CLI agent that streams to its terminal — no per-provider integration required.