Daybook

Honest daily reflection — track your energy and output over time, without the noise.

Live app: daybookhere.vercel.app

Daybook is a small, focused habit-tracking app. Each day you log how much energy you had, how much you got done, and whether you stuck to your routine. Over time, it surfaces plain-language patterns from that data, no streaks to game, just a clear record.

Features

• Daily check-in : Rate energy and output (1–5), note whether today matched your routine, and add an optional comment. One check-in per day.
• Routine lock : Set your anchor tasks for the week; once saved, the list locks for 7 days so you can't second-guess it mid-week.
• History : Browse past check-ins, most recent first.
• Patterns : AI-generated insights (via the Claude API) that read your check-in history and surface a plain-English observation, once you have enough data logged.
• Settings : See routine lock status and clear your local data.
• Local-first storage : Check-ins and routines are stored in localStorage; nothing leaves your browser except the anonymized data sent to generate a Patterns insight.

Tech stack

• Next.js 16 (App Router, Turbopack)
• React 19 + TypeScript
• Tailwind CSS 4
• Vitest + Testing Library for unit/component tests
• Anthropic API for the Patterns insight endpoint
• Deployed on Vercel

Getting started

Prerequisites

• Node.js 18.18+
• An Anthropic API key if you want the Patterns feature to work locally

Setup

git clone https://github.com/waleeja07-wk/daybook.git
cd daybook
npm install

Create a .env.local file in the project root:

ANTHROPIC_API_KEY=your-api-key-here
# Optional — defaults to claude-sonnet-4-5
ANTHROPIC_MODEL=claude-sonnet-4-5

Run the dev server:

npm run dev

Open http://localhost:3000.

Testing

npm run test        # run once
npm run test:watch  # watch mode

6/6 tests passing across src/lib/storage.test.ts and components/CheckInForm.test.tsx.

Note on test output: if you see a SyntaxError printed in the terminal while tests run, that's expected — it's console.error output from storage.test.ts's "returns an empty list when stored check-ins are invalid JSON" test. It intentionally writes malformed JSON into localStorage and asserts that getAllCheckIns() catches the parse failure and falls back to [] instead of crashing the app. The test still passes; the logged error is evidence the error handling works, not a failure.

Build

npm run build
npm run start

Project structure

app/
  page.tsx           # Today — daily check-in form / confirmation
  routine/           # Weekly routine editor (7-day lock)
  history/           # Past check-ins
  patterns/          # AI-generated insights
  settings/          # Routine status + data controls
  api/insights/      # Route that calls the Anthropic API
components/          # UI components (forms, panels, nav)
src/lib/storage.ts   # localStorage read/write helpers
src/types/           # Shared CheckIn / Routine types

How data works

All check-in and routine data lives in the browser's localStorage — there's no database and no account system. The /api/insights route accepts your check-in history, forwards it to Claude to generate a short pattern observation, and returns that text; it does not persist anything server-side.

AI integration

The Patterns page calls POST /api/insights with the user's recent check-ins (date, energy level, output level, routine match, optional note). The route validates the payload server-side, requires at least 3 check-ins before calling out to Claude (MIN_CHECK_INS), and asks the model for a short, plain-English observation about the data — not advice, not a score, just a pattern worth noticing.

Failure handling: if the API key is missing, the upstream call fails, or Claude returns something that doesn't match the expected shape, the route returns a typed error (MISSING_API_KEY, UPSTREAM_ERROR, INVALID_RESPONSE, etc.) rather than a raw 500, and the Patterns page shows a calm fallback message instead of crashing.

Deployment checklist

• Environment variables: ANTHROPIC_API_KEY (and optionally ANTHROPIC_MODEL) must be set in Vercel → Project Settings → Environment Variables. They are read server-side only and never exposed to the client.
• Secrets hygiene: .env.local is git-ignored and has never been committed (verified with git log --all -- .env.local).
• AI failure mode: if /api/insights fails for any reason (bad/missing key, upstream error, malformed response), the Patterns page falls back to a plain "couldn't generate an insight right now" message rather than an unhandled error.
• Rollback plan: revert the last commit on main (git revert <sha> or git reset --hard <sha> + force-push) — Vercel auto-redeploys on push to main.

Known limitations / future improvements

• Data is local to a single browser — no sync across devices, no accounts.
• Patterns insights require a minimum of 3 check-ins before they'll generate.
• No offline queueing for the insights call; a failed request simply shows the fallback state until retried.

License

MIT — see LICENSE.