seq

Parse Codex JSONL session logs to extract skill usage and token counts.

66|2|Updated Feb 18, 2015

Search Tags:#reports #skill usage #Codex #sessions #analyze #token counts

Ready to run

npx skills add https://github.com/tkersey/dotfiles --skill seq

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: seq
Source: https://github.com/tkersey/dotfiles/tree/main/codex/skills/seq
Command: npx skills add https://github.com/tkersey/dotfiles --skill seq

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill automates parsing Codex JSONL session logs to extract skill mentions and token usage, enabling quick audits and reporting without manual crunching.

Core Features & Use Cases

Extracts skill mentions from Codex sessions both via <skill> blocks and $skill tokens to build usage datasets.
Computes per-session and per-time-period summaries of token counts to support trend analysis and reporting.
Use Case: Rank skills across a corpus of sessions and generate a table of the most frequent skills with counts.

Quick Start

uv run scripts/seq.py skills-rank --root ~/.codex/sessions

Frequently Asked Questions about seq

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I extract skill usage data from Codex session logs?▼

You can analyze Codex session JSONL logs to extract skill mentions and token_count events. The script parses local log files to surface usage events and compute per-session summaries for reporting.

Can I rank skills by frequency across multiple Codex sessions?▼

Yes, you can rank skills by frequency across multiple Codex sessions. The script identifies skill mentions via blocks and dollar tokens, then generates a table of the most frequent skills with their respective counts.

How do I compute token count trends from JSONL logs over time?▼

You compute token count trends by parsing token_count events from Codex JSONL logs. The script aggregates these events into per-time-period summaries, enabling time-based trend analysis and reporting from local session files.

Do I need external services or dependencies to parse Codex session logs?▼

No, you do not need external services or dependencies to parse Codex session logs. The solution relies solely on the Python standard library and local file access to extract skill mentions and token counts.

What is the best way to audit skill usage in Codex sessions?▼

The best way to audit skill usage in Codex sessions is to automate parsing JSONL logs to extract skill mentions and token usage. This provides quick audits and reporting without manual data crunching.

Does the script support role-based summaries for skill usage events?▼

Yes, the script supports role-based summaries for skill usage events. It surfaces skill mentions and token_count events from Codex logs to enable both time-based trends and role-based reporting.