What I learned building a local monitor for Codex and Claude Code sessions

I use both Codex and Claude Code and I wanted a better way to understand what was happening locally without jumping between logs and different tools. While building a small dashboard for this, I learned a few interesting things. Codex stores session data locally under ~/.codex/sessions, while Claude Code keeps project/session data under ~/.claude/projects. That meant I could build the monitor without sending session data to an external service. The dashboard runs a local Node.js server bound to 127.0.0.1 and reads the local session files to extract things like recent activity, token usage and session information. One of the harder parts was turning raw usage data into something actually useful. For example, instead of only showing a quota percentage, I started keeping local history so the dashboard can show how quickly quota is being consumed over time. I also ended up making two views. Easy mode is meant to answer: "Am I okay to keep using the agent?" Advanced mode exposes more of the raw session and quota information. Another decision I made was not to display or upload prompt content. I only wanted the monitoring data needed for the dashboard. I'm still experimenting with what information is actually useful when you're running coding agents for several hours. For people using Codex or Claude Code regularly, what would you actually want to see in a dashboard like this?