01 · Roasts
63% PLpgSQL and counting
Your language breakdown is 63% PLpgSQL. You have a Next.js app, an Android app, a satellite AI system, and your profile is screaming 'I am a database'. Nobody put 'writes migrations for fun' on their bio.
5-minute hackathon ghost
hackagent was created and abandoned within 5 minutes — 2 commits, 23KB, enrichEvent() cut off mid-function. You didn't even finish the sentence.
191 commits, mostly Fridays
191 public commits in a year and your heatmap looks like a morse code signal with long silences. Weeks 5–13 are basically a ghost town. The burst in week 44 (all 4s) then silence suggests the GitHub graph is powered by hackathon adrenaline.
vulnshop is self-aware
You literally wrote 'vibecoded slop' in the repo description and still pushed it public. Respect the honesty. Zero points for the code.
143 stars, 0 tests, no CI
detour won a TreeHacks honorable mention and has 143 stars — impressive. It also has zero tests and no CI. One day someone will try to run it and discover the orbital physics only works on your laptop.
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% weight58D
- Consistency20% weight60C
- Quality20% weight67C
- Depth15% weight65C
- Breadth10% weight72B
- Community10% weight50D
03 · Stats
365-day commit heatmap
92 active days
Language distribution
- PLpgSQL63%
- TypeScript10%
- Kotlin8%
- Go7%
- HTML3%
- Python3%
- Other6%
04 · Numbers
Owned repos
non-fork
33
Commits
last 12 months
191
Followers
124
Joined GitHub
Nov 2013
05 · Top repos
keanucz /
detour
TreeHacks 2026 honorable mention project: multi-agent LLM collision-avoidance pipeline for satellites using Nemotron on NVIDIA GX10. Typed TS/Python, structured with physics engine + agents + Next.js UI. Notable scope but nascent ecosystem.
keanucz /
cook-dis
Accessible recipe app combining URL scraping, AI image generation (Runware), and TTS narration for people with dyslexia. Typed Next.js 16 + Supabase with structured multi-service architecture and background job queue, but early-stage with 0 stars, minimal adoption signal, and no tests/CI.
keanucz /
AdobeConnectDL
Go CLI tool for downloading Adobe Connect recordings with video embedding, metadata extraction, and subtitle support. Typed, well-tested, documented, and structured—but niche utility with minimal stars and no external adoption signals.
keanucz /
squish.space
Kotlin/JS interactive 3D blob + math graphing tool built at hackathon with hand gesture tracking, spring physics, and procedural audio. Typed, well-structured, documented, and shipped to production with live demo.
keanucz /
trainstudent
Production-intent Android+Go app for UK train discounts with Kotlin+Compose frontend, voice NLU (Cactus LLM), and backend integrating Northern Railway, Avanti, Grand Central APIs. Typed, documented, structured, but no tests/CI and very recent creation (Jan 2026)
keanucz /
voicesum
Early-stage React Native voice summarization mobile app with polished UI components but minimal documentation, no tests/CI, and only mock data—experimental prototype stage with professional styling but unproven functionality.
keanucz /
keanuc.net
Personal portfolio/blog site built with Zola static site generator. Minimal scope with 2 stars, 12 commits in 4 months, typed-language requirement unmet. Ships with MIT license, CI, and gitignore but lacks tests and meaningful architectural documentation beyond README.
keanucz /
activity-logger
Personal WhatsApp bot that transcribes audio to Google Sheets using Whisper + ChatGPT. Single file (main.py), minimal dependencies, no tests/CI, untyped Python with basic documentation.
keanucz /
hackagent
Hackathon bot built in Go for Encode AI Hackathon 2026 competition. 23KB single-file codebase with feature roadmap (scraping, reminders, team matching) but incomplete implementation and no tests/CI. Created and last pushed same day (2026-03-22), abandoning after 2 commits.
keanucz /
vulnshop
Empty scaffold with no documentation, tests, or CI. 29KB JavaScript project with 1 commit in 36 days, explicitly described as "vibecoded slop" for a hackathon test.
06 · Timeline
- Nov 17, 2013Joined GitHub
- Jan 1, 2025Created activity-logger — transcribes whatsapp voice notes into rows on a google sheet
- Jun 28, 2025Created voicesum
- Dec 8, 2025Created AdobeConnectDL — Utility that enables you to download one or more Adobe Connect recordings & associated metadata
- Dec 11, 2025Created keanuc.net — my portfolio website + blog
- Jan 25, 2026Created cook-dis — cook dis: for people with dyslexia or otherwise who just want to make any long online recipe into a step-by-step thing
- Jan 31, 2026Created trainstudent — save as much money as possible on train tickets with random student discounts and quickly assess train info when you need it
- Feb 14, 2026Created detour — On-board AI agents autonomously saving satellites from orbital debris @ treehacks 2026
- Feb 22, 2026Created vulnshop — vibecoded slop to test my hackathon project
- Mar 22, 2026Created hackagent
- Apr 1, 2026Created squish.space — squish 3D graphs with your hands
- Apr 21, 2026Most recent push to activity-logger
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.