ProjectsCase Study

Author · 2026

Codex Usage Tracker

A local tool for importing Codex usage logs into SQLite and exploring them through a CLI and dashboard.

Open-source local tool

PythonSQLiteNext.js
Codex Usage Tracker — original portfolio artwork

Why I built it

I wanted to examine coding-agent usage across sessions and projects without repeatedly parsing separate log files.

What it does

Codex Usage Tracker imports supported local logs into SQLite and exposes reports by date, model, directory and session. A Next.js dashboard provides a visual view of the same records, while export commands support further analysis.

Implementation focus

The project combines incremental ingestion, local storage and timezone-aware reporting. Redaction and purge controls are intended to limit the content retained alongside usage metadata.

Interpreting the records

Supported inputs include rollout JSONL, CLI logs and app-server JSON-RPC logs as documented by the source. Reports depend on parser support for the log version. Input, cached-input and output fields must be interpreted according to that format before combining totals. This page publishes no personal usage totals, session text or price estimates.