Work Log

ADHD-friendly work tracker — single HTML file, runs locally in your browser.

Screenshots

How to use

Important: The tracker needs to run via a local server (not opened directly as a file) so your data saves correctly between sessions.

Windows

1. Download work-log.html, launch.bat, and start-server.ps1 into the same folder
2. Double-click launch.bat
3. The tracker opens automatically in your browser

Linux / Mac

1. Download work-log.html and launch.sh into the same folder
2. Open a terminal in that folder and run:

   bash launch.sh
   

3. The tracker opens automatically in your browser

Your data is stored locally in your browser — nothing is sent anywhere.

Data lives per-origin, so keep the port stable. Your entries are stored in your browser's localStorage, which is scoped to the exact origin:port you loaded the page from. Both launchers use a fixed port (8080) for this reason — if you ever load the tracker from a different port, your previous data won't show up (it's not gone, it's just stranded on the old port). Reload http://localhost:8080/work-log.html to get it back.

Features

Time tracking

• Live timer — start a timer on any task, pause and resume, add a handoff note when stopping
• Billable tracking — mark epics and individual entries as billable; totals shown per task and per day
• Quick pick — recent tasks shown as chips for fast re-logging
• Long-running timer warning — dismissible alert once a running timer passes 4 hours, with a one-click stop; also asks whether to stop if you return to the tab after it's sat hidden 30+ minutes mid-session

Tasks

• Kanban board — drag tasks between To Do / In Progress / Done columns with epic colour coding
• WIP warning — amber highlight and dismissable banner when more than one task is in progress
• Task subtasks — split any task into child steps; completing all steps completes the parent
• Deadlines — date picker with overdue (red) and due-today (amber) highlighting
• Auto-carry — unfinished tasks roll over to the next day automatically
• Jira import — paste a Jira CSV export to bulk-add tickets as tasks
• Weekly plan review — a dismissible checklist appears once a new week begins for any task planned ahead ("upcoming") and dated within it, so you can confirm it's still accurate before the week starts

Today's Flow

• Flow view — chronological feed merging time entries, log notes, and task updates for the day
• Log view — editable time log with duration, epic, billable flag, and ad-hoc entry row
• Proof links & notes — attach a link (Confluence page, Zephyr key, filename, or a URL) and a one-line note to any entry; shown as 🔗/📝 indicators and carried into exports
• Blocks view — visual 08:00–18:00 timeblock grid auto-filled from logged entries; drag to rearrange
• Month view — heatmap of hours per day with intensity colouring; monthly summary and task inventory
• Rolling Summary — AI-ready summary of recent activity across configurable days

Focus mode

• Focus screen — one-click distraction-free view showing only the active task and next steps
• Parked thoughts — capture stray thoughts mid-focus without leaving the screen
• Pomodoro timer — 5 / 10 / 20 min ring timer with session log; visible in focus mode

Planning

• Today's meetings — fetched live from Outlook calendar (Windows); shows time, duration, Teams join link. Only your own personal calendars are read — your mailbox and its archive, including secondary calendars nested inside another folder — not shared/delegate mailboxes or public folders, which are excluded automatically. A meeting counts as today's if it overlaps today at all — so multi-day events and meetings that run past midnight appear too. ✕ hides a single meeting for the rest of the day, not every meeting sharing its title. If long-running recurring meetings are missing, raise $CalendarLookBackYears in config.local.ps1 (see config.local.example.ps1). If a calendar you don't want still shows up — e.g. a shared calendar that reports as an ordinary folder — name it in $CalendarExcludeNames. http://localhost:8080/api/calendar?debug=1 reports what the collector found.
• Day streak — consecutive days with logged work

Export & review

• End the day — one-click summary with test areas and tomorrow's notes, exported as .txt
• Daily hours entry draft — the End of Day modal shows a draft of the day's hours entry: total hours plus an activity (ticket, ticket) description built from your log. Edit it and press "copy description" to paste into your time-tracking system. One-click submit (Claude in Chrome first, Playwright as fallback) is built but switched off: set EOD_SUBMIT_ENABLED = true in src/js/00-config.local.js to show the submit button, and $EodSubmitEnabled = $true in config.local.ps1 to let the server accept it. It needs start-server.ps1, the Claude Code CLI with the Claude in Chrome extension, and/or a one-time npm run eod:login (Microsoft SSO) for the Playwright fallback, whose page selectors are unverified defaults — override them in scripts/eod-submit.selectors.json.
• Auto-backup — JSON backup saved automatically on end-of-day to a local JSON backups/ folder
• Weekly report — copy-to-clipboard draft summarising the calendar week's tracked time grouped by Jira ticket, for writing status updates without reconstructing "what did I touch" by hand
• Gap report — flags this week's finished, billable entries that are missing both a note and a proof link, with a one-click jump to fix each one

Bullet Journal (BuJo) features

• Rapid logging — ✏️ button opens a floating capture panel; Enter starts the timer immediately
• Signifiers — clickable symbol on each entry cycles through meeting / flagged / migrated / cancelled / overtime; cancelled entries are excluded from totals
• Daily Log — tab view merging time entries, log notes, and task updates in a single chronological feed
• Monthly Log — heatmap of hours per day with intensity colouring; monthly summary and task inventory
• Migration — end-of-month close-out flow: carry forward, reschedule, or drop each open task
• Trackers — custom 28-day progress grids with a daily target and streak counter
• Sprints — intention-first focus sessions: declare an outcome, run a Pomodoro, then record yes / partly / no

Info widgets

• Weather — current conditions, rain forecast, sunrise/sunset (Helsinki)
• Moon phase — current phase, illumination %, and zodiac sign
• Finnish nameday — fetched live from nimipaivat.fi

Hours submission — security model

• Off by default. EOD_SUBMIT_ENABLED is false in src/js/00-config.js; nothing is sent until you set it to true in your own src/js/00-config.local.js and $EodSubmitEnabled = $true in config.local.ps1. The browser flag only shows the submit button; the server enforces its own switch and answers POST /api/eod-submit with 403 while $EodSubmitEnabled is not a real $true (the default).
• Cross-origin requests are refused. start-server.ps1 no longer sends Access-Control-Allow-Origin: *, and the endpoint rejects any request whose Origin is not this server's own http://localhost|127.0.0.1|[::1]:<port>, or whose Host is not a local name (DNS-rebinding), with 403. A web page open in your browser therefore cannot trigger a submission through your signed-in session. Requests with no Origin (e.g. curl on your own machine) are still accepted when submission is enabled.
• Everything runs on your machine. There is no remote service of ours. scripts/eod-submit.mjs is started by the local server and drives your signed-in Chrome (claude --chrome) or a Playwright profile against $EodSubmitUrl. The date, hours and description go over stdin, never argv. The Chrome route runs the Claude Code CLI, so that entry text is also seen by Claude (Anthropic API); the Playwright route does not involve it.
• No credentials in the repo. Sign-in is Microsoft SSO done by hand (npm run eod:login); the session cookies live in .timesheet-profile/. scripts/eod-submit.selectors.json holds only optional CSS selector overrides ({} by default).
• Gitignored local files: config.local.ps1, src/js/00-config.local.js, .timesheet-profile/. Copy config.local.example.ps1 / src/js/00-config.local.example.js and keep the copies untracked.

Troubleshooting

My entries are gone after reopening the page. Entries live in localStorage, scoped to the exact origin:port. Reload http://localhost:8080/work-log.html (see the note under How to use). If you still see nothing, check that your browser isn't set to clear site data on exit, then restore from the newest file in JSON backups/ (written automatically at end-of-day).

The timer didn't save. Same cause as above — a different port or cleared site data. Log the missed time manually from the Log tab.

Project documentation

Document	Purpose
ARCHITECTURE.md	Module map, data flow, and design decisions
DATA.md	localStorage schema dictionary
CONTRIBUTING.md	Dev setup, branching strategy, and quality standard
ROADMAP.md	Planned and in-progress features
CHANGELOG.md	Full release history
QA.md	QA checklist and stakeholder sign-offs per release
JSDoc reference	Generated API docs for every exported function, rebuilt on every push to main

Testing

A smoke test suite is included. Requires Node.js ≥ 22.22.1 (matches engines in package.json — the actual floor imposed by lint-staged's own minimum).

node smoke-tests.cjs

Or double-click run-tests.bat on Windows.

If your environment has a pre-provisioned Chromium binary at a revision Playwright's own installer can't reach (e.g. an offline sandbox), point the smoke tests at it with PLAYWRIGHT_CHROMIUM_EXECUTABLE_PATH=/path/to/chromium node smoke-tests.cjs instead of running npx playwright install.

To schedule tests to run automatically each morning, run schedule-tests.bat once as Administrator.

To set up automated weekly releases every Friday, run setup-scheduler.ps1 once as Administrator.