[FEATURE] Usage monitoring dashboard with timestamped session/action correlation

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

When users experience unexpected high usage, there's no way to diagnose what caused the spike. The current usage tracking doesn't provide visibility into:

• When usage occurred (timestamps)
• Which session was responsible
• What specific actions consumed tokens
• The relative cost impact of model selection

Proposed Solution

A usage monitoring dashboard/graph that provides:

1. Timestamped usage graph - Visual timeline showing token consumption over time (hourly/daily views)
2. Session correlation - Each data point linked to the session ID that generated it, allowing users to click through and see what was happening
3. Action-level breakdown - Show which operations consumed tokens:

• Tool calls (Read, Glob, Grep, Task agents, etc.)
• Model responses
• Agent spawns (especially subagents)

1. Model-aware visualization - Normalize or highlight cost by model:

• An Opus 4.5 search showing as a large spike
• The same search with Haiku showing as a small blip
• Makes it clear why certain operations were expensive

Alternative Solutions

_No response_

Priority

Medium - Would be very helpful

Feature Category

Interactive mode (TUI)

Use Case Example

• "Why did I use 500K tokens yesterday?" → See a spike at 3pm, click to see it was a Task agent running Opus doing extensive code search
• "Which model should I use for exploration?" → Compare historical usage patterns between Opus/Sonnet/Haiku for similar tasks

Additional Context

This would help users:

• Debug unexpected usage spikes
• Make informed decisions about model selection
• Understand the token cost of different workflows
• Optimize their usage patterns