01 · Roasts
April Avalanche
You created beyond-tty, slim-ml, glm-code, kaiser, Pelgo-career, and the GSoC placeholder all within 16 days of April 2026. That's 6 repos in half a month — impressive output or evidence that none of them will see May?
The Graveyard Gardener
63% stale repo ratio on 112 public repos means ~70 repos are quietly rotting. Your GitHub profile is less a portfolio and more a archaeological dig site of abandoned ideas.
Test-Allergic Architect
beyond-tty has 10 crates, a CLAUDE.md 400 lines deep, and GPU rendering — but HAS_TESTS=no. You'll document the renderer.rs cache invalidation strategy but won't write a single assert. Respect.
The 40-Minute Shipper
karp was created AND fully committed between 18:09 and 18:47 on April 4th — 38 minutes from zero to npm publish. Either you type 200 LOC per minute or this was a very well-rehearsed idea.
11 Stars, 112 Repos
That's 0.098 stars per repo. You are statistically more likely to get a star by asking a friend than by shipping another experimental LLM tool. The market has spoken, softly.
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% weight62C
- Consistency20% weight65C
- Quality20% weight57D
- Depth15% weight58D
- Breadth10% weight70B
- Community10% weight55D
03 · Stats
365-day commit heatmap
208 active days
Language distribution
- Python58%
- TypeScript26%
- C++8%
- Jupyter Notebook5%
- Makefile1%
- Rust1%
- Other1%
04 · Numbers
Owned repos
non-fork
71
Commits
last 12 months
249
Followers
22
Joined GitHub
Aug 2021
05 · Top repos
PewterZz /
beyond-tty
Early-phase AI-native terminal in Rust with GPU rendering (wgpu+glyphon), block-oriented data model, agent supervision, and capability-based security. Working MVP with solid architecture but nascent adoption (2 stars, <30 days old).
PewterZz /
slim-ml
Early-stage LLM inference optimization framework targeting constrained hardware. Typed Python, well-documented (design.md, ARCHITECTURE.md, docs/), structured src/ layout with tests and CI. Non-trivial architecture for speculative decoding and adaptive memory budgeting, but young (created 2026-04-18, 19 of last 30 comm
PewterZz /
Pelgo-career
Pelgo is an AI-first career intelligence platform that automates job-candidate matching via multi-step LLM orchestration. Features typed Python/Pydantic schemas, FastAPI+SQLAlchemy backend, Docker/Postgres, and comprehensive agent tooling, but created <4 days ago with 19 commits—early-stage experimental project.
PewterZz /
glm-code
TypeScript fork of Claude Code CLI with telemetry removed and features unlocked. Well-documented (README, ARCHITECTURE.md, STATUS.md, FEATURES.md) and recently active (30 commits in 4 days), but 0 stars, no tests/CI, no license, and early-stage (created 2026-04-14).
PewterZz /
clawd-assist
Specialized Android overlay app for OpenClaw AI gateway on e-ink tablets; functional Kotlin project with typed code, structured layout, meaningful README, but minimal adoption signals and recent burst development.
PewterZz /
kaiser
A TypeScript-based AI study tool that bridges the Anthropic Claude CLI to a local Next.js web UI with math rendering, plotting, Python execution, and diagram support. Typed, well-documented, but very fresh (created 2026-04-27, single commit visible) with no tests or CI—classic solo experimental project.
PewterZz /
pstack
Personal reference cookbook CLI for injecting context into local LLMs. Ships typed Node.js code with CLI, setup automation, and 29+ skill directories. Created and last pushed within 24 hours (Apr 5, 2026). Meaningful README + CLAUDE.md docs present. Lacks tests, CI, and license file.
PewterZz /
karp
Nascent CLI tool for local LLM command recall with BM25 retrieval, clean README and functional core, but no tests, CI, type safety, or production usage yet. Created and pushed same day (2026-04-04).
PewterZz /
slice-of-work-mc
Minecraft Fabric mod integrating screen capture and VNC remote display into in-game monitor blocks. One-week dev sprint with functional architecture but no documentation, tests, or CI pipeline.
PewterZz /
homebrew-tap
A nascent Homebrew tap providing single formula distribution with minimal project structure, no tests/CI, and only 3 commits in past 30 days suggesting one-off setup.
PewterZz /
PewterZz
Profile config repo with minimal scope—GitHub README with badges and links only. No source code, tests, CI, or meaningful project substance. 41 KB total size with recent activity but no architectural depth.
PewterZz /
Gsoc-gemini-cli-proposal
Empty scaffold repo created 2026-04-15 with 1 commit, no documentation, tests, CI, or meaningful code content. Appears to be an initial placeholder for a Google Summer of Code proposal project.
06 · Timeline
- Aug 3, 2021Joined GitHub
- Aug 3, 2021Created PewterZz — Config files for my GitHub profile.
- Mar 17, 2026Created clawd-assist — Floating Android chat overlay that connects to a local OpenClaw gateway. Built for e-ink tablets.
- Mar 30, 2026Created slice-of-work-mc
- Apr 4, 2026Created karp — cli-tool for using local llms to grab commands/notes quick
- Apr 5, 2026Created pstack — My oddly specific opinionated toolset for injecting context into gemma-4-e2b-it-q5_k_m/Qwen3.5 35B A3B Opus distills for higher quality outputs
- Apr 14, 2026Created glm-code
- Apr 15, 2026Created beyond-tty — Rust-based AI Native Terminal Emulator
- Apr 15, 2026Created Gsoc-gemini-cli-proposal
- Apr 16, 2026Created homebrew-tap — Homebrew formulae by Peter Subrata
- Apr 18, 2026Created slim-ml — An adaptive LLM runtime for running medium models on constrained hardware.
- Apr 27, 2026Created kaiser — ai study tool
- May 1, 2026Created Pelgo-career
- May 1, 2026Most recent push to Pelgo-career
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.