STARlog helps you capture, structure, and rehearse past-experience stories using the STAR format (Situation, Task, Action, Result). Speak or paste a rough description of what happened — Gemini AI turns it into a polished, interview-ready story. Map your stories to specific job postings, see coverage at a glance, and walk into any interview with the right story ready to go.
Everything stays in your browser. No account, no server, no data leaving your machine — except the Gemini API calls made with your own key.
- Job-first navigation — each job you're interviewing for lives in the sidebar. The active job's hub shows coverage at a glance.
- Job hub — paste a job description; Gemini extracts the 5–7 behavioural competencies the role is likely to interview on. See which are covered, which are gaps, and map existing stories to any competency in one click.
- Gap fill — click + draft on any uncovered competency to launch a focused capture flow pre-tagged with that competency. Gemini extracts and maps the story automatically on save.
- Capture your way — record audio directly in the browser, upload an existing file, or paste a transcript. Gemini extracts a structured STAR story from any of them.
- Story bank — searchable table of all your stories. Text search covers title, competency tags, situation, and result. Click any row to open the editor.
- Story editor — edit any STAR field inline, manage action steps, add private notes.
- Interview mode — full-screen prep view with a launch pad offering three modes: Flash cards (keyboard-navigable card deck, ← → between stories, ↑ ↓ between competency groups), Mock interview (question-first, space to reveal), and Drill with timer (90 s target, 1–5 self-rating). Two-line action crib visible by default; click/space to expand the full STAR.
- API key validation — live ping on the key you enter; Save is disabled until the key is confirmed working.
- Node.js 22+ — nodejs.org
- Gemini API key — free tier available at aistudio.google.com/app/apikey
git clone https://github.com/stefanhoth/starlog.git
cd starlog
npm install
npm run devOpen http://localhost:5173, enter your Gemini API key, and start capturing stories.
| Layer | Choice | Why |
|---|---|---|
| UI framework | Svelte 5 (runes mode) | Fine-grained reactivity without a VDOM; native browser APIs fit naturally |
| Styling | Tailwind CSS v3 + DaisyUI v4 | Utility-first layout, component tokens for consistent UI |
| Build | Vite 6 | Near-instant HMR in development; fast production builds |
| AI | Gemini 2.5 Flash (@google/generative-ai) |
Multimodal (audio + text in one call); generous free tier |
| Storage | IndexedDB (idb) for stories, profiles, and settings; one-time migration from localStorage on first access; sessionStorage for transient UI state |
No backend needed; data never leaves the device |
| Routing | Hash-based (#/) via History API |
Deep links and back/forward work on GitHub Pages without server config |
| Tests | Playwright (Chromium) | Real browser, real MediaRecorder, Gemini mocked via page.route |
Story {
id, title, original_language
star: { situation, task, action: string[], result }
quality: { situation, task, action, result, notes } ← AI-assessed per section
competency_tags, rank, notes
}
JobProfile {
id, company, role, jobDescription
extractedCompetencies: string[]
competencyMap: Record<competency, storyId[]> ← same story usable across profiles
}
npm run dev # dev server with HMR at localhost:5173
npm run check # svelte-check + tsc (type errors)
npm run build # production build to dist/
npx playwright test # full e2e suite (55 tests, Chromium)
npx playwright test tests/onboarding.spec.ts # single fileTest fixtures live in tests/fixtures/. Gemini responses are mocked — no real API key needed to run the suite.
src/
lib/
components/ # StoryMapModal (multi-select), StoryCard, JobProfileCard
stores/ # stories, jobProfiles, settings, view (IndexedDB-backed)
gemini.ts # extractSTAR, extractCompetencies, verifyApiKey
audio.ts # AudioRecorder (MediaRecorder wrapper)
competencies.ts # preset competency tag list
types.ts # Story, JobProfile, Settings, StoryDraft
views/
App.svelte # layout shell + view router (job-first sidebar)
Onboarding.svelte # key setup → add-first-job → review competencies
JobHub.svelte # per-job competency coverage + map/draft actions
StoryBank.svelte # full story table with text search
Capture.svelte # record / upload / text → Gemini → Review (gap-fill is a mode of this view)
Review.svelte # edit & save extracted STAR draft
StoryDetail.svelte # inline STAR editor, rank, tags, delete
InterviewMode.svelte # full-screen prep: flash cards / mock / timer modes
tests/
fixtures/ # star-draft.json, competencies.json
helpers.ts # shared Gemini mock helper
*.spec.ts # one spec file per view
GitHub Actions runs on every push and PR to main:
| Job | Trigger | Steps |
|---|---|---|
| Quality | push + PR | npm run check → npm run build → npm audit |
| E2E | PR only | Playwright tests in pre-baked Chromium container |
| Deploy | push to main | Build → publish to GitHub Pages → CalVer release tag |
Every successful deploy creates a YYYY.MM.DD release tag (e.g. 2026.05.20). Multiple deploys on the same day get a numeric suffix (.1, .2, …). See CHANGELOG.md for a human-maintained history of notable changes.
Dependency updates are managed by Renovate (weekly, Monday mornings CET). Dev dependency patches and minor versions automerge when CI passes. The Gemini SDK is flagged for manual review.
GNU General Public License v3.0 or later (GPL-3.0-or-later). STARlog is free and open source — you're welcome to read, use, and modify the code under the terms of the GPL.