adunne09/deepwork-tracker
Overview
This skill tracks local deep work sessions, letting you start, stop, check status, and generate shareable reports. It runs a SQLite-backed local script at ~/clawd/deepwork/deepwork.js and produces text or Telegram-ready heatmaps in a GitHub-contribution-graph style. Use it to capture focused time locally and share summaries quickly.
How this skill works
The skill executes the local deepwork script to start and stop timed sessions, optionally launching a macOS Clock timer for the target duration. It inspects the script's status output to report active sessions and reads the SQLite data when generating reports or heatmaps. For sharing, it can format a Telegram-ready report and send it to the configured recipient.
When to use it
- Start a focused session: “start deep work” or specify a target duration.
- End a session and get the duration: “stop deep work.”
- Check whether you are currently in a session: “am I in a session?”
- Share recent activity as a text summary or a Telegram-ready heatmap: “show my deep work graph.”
- Review session history over ranges like 7, 14, 30, or 60 days.
Best practices
- Keep the local script at ~/clawd/deepwork/deepwork.js and executable for reliable behavior.
- Specify a target duration when starting (default is 60 minutes) to get consistent timers.
- Grant macOS Accessibility permissions if you want automatic Clock timer control.
- Use the --days or --mode flags to control report range and output format explicitly.
- When sharing via Telegram, expect the skill to send the report to the configured Telegram ID.
Example use cases
- Start a 90-minute block: run start with --target-min 90 and get a confirmation.
- Stop a session and receive the actual focused duration in reply.
- Request a 7-day text summary of minutes per day and review recent trends.
- Generate a 52-week heatmap and send it to Telegram for weekly accountability.
- Ask for status mid-work to confirm the active session and remaining time.
FAQ
I will report missing pieces and still attempt available actions; you can bootstrap the script into ~/clawd/deepwork/deepwork.js to restore full functionality.
Who receives the Telegram report?
By default the skill sends Telegram-ready output to the configured recipient (the workflow sends to the specified Telegram ID for sharing).