Score breakdown
🧬 Most similar developers
Closest profile, nearby score
🔥 Full roast
🔥 Your own repos top out at 208 stars, half your recent external PRs are bulk templated task spam for a 42-star repo, you closed 6 of your own external PRs, and even your profile config repo has 0 stars
Leolty — 71.30/100 · SOLID (Solid · Trustworthy)
TL;DR: Solid LLM research contributor with legitimate merged PRs to high-star reasoning and benchmarking repos, but his own project portfolio is anemic and recent external work leans heavily into low-effort templated task filler.
| Dimension | Score | Notes |
|---|---|---|
| Account maturity | 10/10 | Registered 6.59 years, active across 8 contribution years, last activity 107 days ago — long tenure, but recent output leans toward low-effort bulk adds. |
| Original project quality | 9.5/18 | Total stars 242, top original repo is repobench (208 stars, ICLR 2024 benchmark, quality score 0.79); the other 15 repos are mostly duds, including a 0-star profile config repo and multiple 0-1 star notebooks and toy projects. |
| Contribution quality | 19/27 | 24 merged PRs out of 35 total; 0 maintainer-rejected PRs, 6 author-closed external PRs. Recent merged sample includes 5 trivial PRs and 13 doc-like/task-add PRs, with half of recent external PRs targeting the tiny 42-star CocoaBench repo with copy-paste templated task content. |
| Ecosystem / maintenance impact | 17.3/20 | 15 all-time substantial PRs/commits into popular repos, including 95 commits and 9 PRs to maitrix-org/llm-reasoners (2,344 stars) and 13 commits to Ber666/ToolkenGPT (271 stars); the bulk CocoaBench task work is correctly categorized as low-value benchmark filler, not core engineering. |
| Community influence | 5.7/8 | 60 followers, 9 following — a tiny audience for a 6-year account with a UCSD CS PhD, no community pull beyond direct project contributions. |
| Activity authenticity | 9.8/17 | 192 contributions last year, no bot or farming flags, but recent output is heavily skewed toward bulk templated task additions to small external repos rather than original engineering work. |
Red flags
- Recent external PRs are dominated by bulk templated task additions to the 42-star cocoabench/cocoa-agent repo, with almost no original engineering value per PR.
- 6 self-closed external PRs, showing a pattern of opening contributions that get abandoned or revised without maintainer merge.
- Own project portfolio is almost entirely low-star or zero-star, with no strong original engineering projects beyond the ICLR 2024 repobench benchmark.
- No contributions in the last 107 days, a recent inactivity streak for a 6-year account.
Score calibration No extra adjustment. The base score already balances legitimate contributions to popular LLM repos against low-value bulk PR work, with no missed qualitative signals warranting a score bump or haircut.
Verdict Normal, solid research-focused contributor with legitimate work in LLM reasoning and benchmarking, but his home turf is surprisingly thin and recent output leans heavily into low-effort templated task spam. No signs of farming or bot activity, but he's nowhere near the impact his PhD bio might suggest.
📌 Pin to your GitHub profile
Copy any snippet below into your Profile README. The score stays up to date automatically, and the image links to your detail page.
[](https://githubroast.dev/u/leolty)<a href="https://githubroast.dev/u/leolty"><img src="https://githubroast.dev/api/badge/leolty" alt="GitHub Roast score badge" /></a>[](https://githubroast.dev/u/leolty)<a href="https://githubroast.dev/u/leolty"><img src="https://githubroast.dev/api/card/leolty" alt="GitHub Roast" width="600" /></a>