Every reading, explained

What Console's cells and bodies, the coach's four lights and each of the nine panels show, what every reading means, and what to actually do when it moves. Each entry says how the reading appears on screen, what it tells you, the change to make, and where the number comes from.

• —HeaderConsole's first row: the session at a glance — its model, the turn, how long it has run, and the phase cell with the turn's elapsed time.
• cCoachThe four lights behind Console's cells — context, cache, limits, rework — and the one nudge the coach keeps on the act line.
• 1ContextHow full the context window is, how fast it's filling, and how many turns are left before the next compaction rewrites it.
• 2Tokens & CostWhere every token went — cache read, cache write, fresh input, output and thinking — and what the session costs, live.
• 3LimitsAccount-wide 5-hour and 7-day usage, when the window resets, and whether the current trend runs out before it does.
• 4TurnWhat the current turn is actually waiting on — the model, a tool, a hook, a permission prompt, or nothing at all.
• 5ToolsCall counts, error rates, latency, and exactly how many tokens each tool pushed into context.
• 6Agents & MCPSubagents and MCP server processes running alongside the main thread, and what each is costing.
• 7FilesEvery file this session touched, how much changed, and which reads were wasted re-reads.
• 8EventsThe chronological stream of tool, hook, permission and compaction events behind every other panel.
• 9AdvisorAll 36 rules the coach checks, the class each fires in, what triggers it, and what to do about it.