01 · Roasts
Star-Commit Ratio Disorder
3,494 commits in a year, 256 PRs, 391 issues... and 30 total stars across 24 repos. You're essentially a one-person open-source factory shipping into a void. The CVEs have more real-world impact than your entire star count.
Homebrew Tap of Shame
You built a Rust tool, a Python AI runtime, and a TypeScript audit CLI — then distributed them via a 1KB Homebrew tap with 0 README, 0 tests, and 2 commits. The packaging is the bug.
5-Day Startup Veteran
unfcked, vibechecked, vex, agent-undo — all shipped within days, all architecturally complete, all with 0–2 stars. You have the velocity of a YC batch and the audience of a private repo.
Test Debt Accumulator
7 repos have HAS_TESTS=no including your 3257KB password manager and your 2543KB student dashboard. You write ARCHITECTURE.md before you write test_*.py. Bold strategy.
The Heatmap Doesn't Lie
Weeks 48–52 are a sea of 4s — you're coding every single day including weekends at 52% night-owl rate. At this pace, you'll have 7 published CVEs and 7 published npm packages before anyone notices either.
Built using
Zoral
Shadows one worker for a week, then takes over their job with zero extra setup. Behaves exactly like the original.
zoral.ai
02 · Category breakdown
- Impact25% weight68C
- Consistency20% weight95S
- Quality20% weight72B
- Depth15% weight60C
- Breadth10% weight80A
- Community10% weight55D
03 · Stats
365-day commit heatmap
309 active days
Language distribution
- Python45%
- TypeScript29%
- Astro8%
- Rust7%
- JavaScript5%
- CSS2%
- Other4%
04 · Numbers
Owned repos
non-fork
16
Commits
last 12 months
3,494
Followers
17
Joined GitHub
Jan 2021
05 · Top repos
peaktwilight /
unfcked
Production-ready TypeScript CLI tool with 45+ code quality checks, comprehensive test coverage, CI/CD, and named domain. Early-stage project (5 days old, 30 commits) shipping real value for vibe-coded projects.
peaktwilight /
whatdiditdo
TypeScript CLI tool for auditing AI-generated code changes with git diff parsing, security scanning, PR generation, and Slack/Discord webhook integration. Well-structured, tested, typed, and shipped with CI/CD.
peaktwilight /
doruk.ch
Personal portfolio built with Astro 6 + React islands, Three.js 3D coin, Framer Motion/GSAP animations. Typed TypeScript, well-structured src/, design.md + ARCHITECTURE.md, 721KB codebase. No tests/CI/license. Single-repo presence, ~30 commits visible, deployed to Vercel.
peaktwilight /
agent-undo
Locally-scoped Rust agent-undo tool with comprehensive docs and strong engineering, but minimal adoption (2 stars, <2 weeks old, fresh burst). Non-trivial codebase that demonstrates serious craftsmanship.
peaktwilight /
infomaniak-cli
Well-structured Python CLI for Infomaniak service management with comprehensive DNS features, tests, CI/CD, and clean modular architecture across 120 KB of code. Published to PyPI (v0.7.0) but early-stage adoption (14 stars).
peaktwilight /
fhnw-dashboard
Modern Next.js 15 student dashboard for FHNW (weather, trains, grades, resources). TypeScript typed, structured multi-file layout with CI, real-time APIs, dark mode, i18n. No license or tests; relatively early-stage with live deployment but limited adoption.
peaktwilight /
vex
Early-stage Python AI workflow CLI (0.1.0) composing uv, PydanticAI, ollama, with sandbox policy enforcement. Typed control plane, comprehensive docs, but nascent codebase with minimal adoption.
peaktwilight /
csv-password-cleaner
TypeScript password management web app with zero-trust client-side architecture. Features smart password organization, security analysis, and generation. MIT-licensed with comprehensive README but no tests or CI; 3257 KB codebase with year-long commit history shows sustained indie development.
peaktwilight /
vibechecked
Newly-launched Astro/TypeScript CLI tool for AI-powered UI design scoring with shareability features. Well-tested and documented but pre-adoption with 0 stars; heavy initial sprint shows capability but lacks sustained trajectory.
peaktwilight /
ThatsAppMQTTDebugger
TypeScript MQTT debugger with live demo and comprehensive documentation. Typed, documented, structured project with CI enabled, but minimal adoption (1 star, no forks). Recent activity and working feature set indicate active portfolio project.
peaktwilight /
peaktwilight
Personal profile/portfolio repository with only a README (truncated HTML/badges) and generated artifact (github-snake.svg). No functional code, no real project deliverables—purely a GitHub profile card.
peaktwilight /
homebrew-tap
Empty Homebrew tap scaffold with no README, tests, CI, or meaningful content. 1KB repo with 2 commits in 7 days shows minimal effort on a placeholder project.
06 · Timeline
- Jan 23, 2021Joined GitHub
- Feb 26, 2024Created doruk.ch — Doruk Tan Ozturk - Official Website
- Mar 15, 2024Created peaktwilight
- Feb 3, 2025Created fhnw-dashboard — A modern, unified dashboard for FHNW students featuring real-time weather, train schedules, mensa menu, and campus resources.
- Mar 7, 2025Created csv-password-cleaner — A secure, privacy-focused tool for cleaning, organizing and generating password exports from Chromium-based browsers.
- May 18, 2025Created ThatsAppMQTTDebugger — A feature-rich MQTT debugging tool for ThatsApp messaging protocol development and testing. Connect to brokers, subscribe to topics, publish messages, and monitor real-time communi
- Feb 27, 2026Created infomaniak-cli — CLI tool for managing Infomaniak services — DNS, mail, hosting, kDrive, and more.
- Mar 29, 2026Created vibechecked — Paste a URL, get a UI roast. Find out if your app is actually good or just another vibe-coded clone.
- Mar 29, 2026Created whatdiditdo — You let the AI cook. Now find out what it actually did. One command after any AI coding session.
- Mar 29, 2026Created unfcked — Your vibe-coded app has problems. This finds all of them. 45+ checks, zero config, no API keys.
- Apr 4, 2026Created homebrew-tap — Homebrew formulae for Peak Twilight tools
- Apr 7, 2026Created agent-undo — Ctrl-Z for AI coding agents. Snapshots every file write, attributes to agents, one-command rollback.
- Apr 17, 2026Created vex — The AI workflow layer for Python — init, dev, eval, deploy in one CLI. Composes uv, PydanticAI, ollama, Inspect AI, Modal/Cloud Run/Docker.
- Apr 22, 2026Most recent push to peaktwilight
07 · Compare
08 · Rubric
How this score was produced
Overall = Σ (category × weight) + gentle top-end curve
Tier thresholds
▸ How the pipeline works
- 01Scrape.Pull every non-fork repo pushed in the last 90 days, plus your contribution calendar, followers, and language byte counts — straight from GitHub's REST & GraphQL APIs.
- 02Triage.A small model reads every repo's file tree + README and picks the 20 files per repo that actually reveal how you code.
- 03Grade each repo. All repos run in parallel through a fast scoring model that reads the picked files and rates each one independently on Impact, Quality, and Depth — with evidence citations.
- 04Aggregate. A larger reasoning model combines the per-repo scores with server-computed stats (heatmap, commit cadence, language entropy, follower count) to produce the 6-dimension profile score + roasts.
- 05Correct.Deterministic server-side checks enforce anchor-scale floors (e.g. a profile with 2,000+ public commits can't score 30 Consistency) and recompute the final verdict.
~90 seconds per profile, ~$0.25 in compute. Total of ~240 files read across your top-12 repos. One rating per GitHub account per day.
▸ Data sources & caveats
- Heatmap & commit totals: GitHub GraphQL
contributionsCollection— covers the last 365 days, includes private repos when the user has opted in (default). - Language %: byte totals across the top 30 owned non-fork repos.
- Curve: a small upward nudge centered on raw score ≈ 70, capping at 100. Prevents specialists from being unfairly penalised for narrow breadth.
- Anchor corrections: when server-measured signals (e.g. privateWorkLikely, multiRepoVolume, follower count) mandate a minimum category score, the aggregation step enforces it. These are signal-conditional, not identity-based floors.