jaobrown/clog
Overview
This skill is a daily coding diary that tracks Claude Code sessions, summarizes work, and optionally shares public leaderboard stats. Use /clog to log recent sessions, /clog init to set up your local repo and config, and /clog status to view aggregated statistics. It automates session discovery, token and duration counting, and pushes logs to your GitHub fork.
How this skill works
On init the skill creates a ~/.clog workspace, config.json and clog.json and (optionally) forks a template repo to your GitHub account. Running /clog scans ~/.claude/projects for sessions since lastSync, extracts summaries, token usage and durations, generates concise one-line summaries when missing, then appends sessions to local logs and pushes changes to your clog repo. /clog status reads clog.json and computes totals, weekly stats, streaks and top projects for a quick report.
When to use it
- After a coding session with Claude to record progress and billing-relevant tokens
- When first installing to configure GitHub fork and local storage (/clog init)
- Before sharing weekly or monthly progress on the leaderboard
- To audit recent token usage and session durations
- When you want a concise daily diary entry pushed to GitHub
Best practices
- Run /clog regularly (daily or after several sessions) so lastSync windows stay small
- Keep ~/.claude and ~/.clog readable by the tool; avoid moving project folders mid-week
- Opt into the leaderboard only if you’re comfortable sharing summary stats publicly
- Keep generated summaries outcome-focused and under ~60 characters
- Pull upstream manually if you edit the repo elsewhere to avoid push conflicts
Example use cases
- Developer tracks Claude-assisted work each day and publishes a monthly log to GitHub
- Engineer audits token spend across projects to identify high-cost sessions
- Team member shares top-project rankings via the public leaderboard topic
- Solo maintainer keeps a searchable timeline of fixes and features for resume or review
- Quickly check current streak and weekly productivity with /clog status
FAQ
/clog will prompt you to run /clog init and will not proceed until config.json and ~/.clog exist.
How are token counts calculated?
The skill sums cache_creation_input_tokens and output_tokens from assistant messages in JSONL, reporting rounded values with k/M suffixes.
Can I opt out of the public leaderboard later?
Yes — set leaderboard:false in ~/.clog/config.json, remove the topic from the repo, then commit and push the change.
