Source Authorship Forensics

Onyx — human-written, or AI-generated?

A source-level read of dylan2554/onyx, contrasted with Stockfish.

Author Dylan (dylan2554)Language C++ (single file)Scope ~1,835 LoC · 21 web-UI commitsEval Self-trained NNUE (own “SABLE” net)Analyzed 2026-07-10Stockfish similarity Moderate →
Verdict · AI-Assisted

Human-directed and self-disclosed — but substantially co-written with AI.

Onyx is refreshingly honest about its own authorship: the engine’s UCI identity literally reports id author Dylan (with Claude). It is a competent single-file C++ engine (magic bitboards, PVS, LMR, PeSTO evaluation) fed by a network the author trained himself, and it descends from the author’s own earlier “Sable” engine (the NNUE format still uses SABLE1/2 magic bytes).

But it lands as finished files via the GitHub web UI — 21 “Add files via upload”/“Delete” commits, no incremental history — so the code was substantially produced with an AI assistant rather than hand-developed in the open. Real human direction (own net, own lineage, open disclosure); heavy AI co-authorship.

Confidence: Medium-High — the self-disclosure is explicit; the exact human/AI split is not.

Part of a series applying the same lens to chess-engine repositories — from the AI-pasted Luna and agent-built Owen, through the AI-assisted Triumviratus, to a range of human-written engines.  See the full index →

1The engine discloses it itself

No forensics needed for the headline — Onyx says so in its own UCI banner.

// onyx.cpp
printf("id author Dylan (with Claude)\n");
Credit where due: openly attributing the AI collaborator in the engine’s identity string is the honest thing to do, and rare. It places Onyx clearly in the AI-assisted band rather than leaving it to be inferred.

2Where the human ends and the assistant begins

#SignalWhat it showsWeight
1UCI id string: author Dylan (with Claude)Self-disclosed AI collaborationNuance
2Own self-trained NNUE (i7-11700K + RTX 3070); descends from the author’s earlier “Sable” engineGenuine human ownership and lineageStrong
3All 21 commits are GitHub web-UI uploads/deletes; single monolithic onyx.cppNo incremental history — code arrives finishedNuance
4Competent, conventional engine (magic bitboards, PVS/LMR, PeSTO eval)Well-formed output — as expected from a good assistantSupport

3Compare & contrast with Stockfish

Where it resembles Stockfish
Bitboards + magic move generation
PVS / alpha-beta, TT, null-move, LMR, killers/history, quiescence
Self-play-trained NNUE evaluation
Aspiration windows
Where it differs
Single 1,835-line file vs a large modular codebase
PeSTO tapered eval as the classical base
Substantially AI-co-written vs decades of human work
Uploaded as finished files, no open dev history

Bottom line

Onyx sits in the AI-assisted middle of the spectrum — and says so itself. A real developer trained his own network, carried forward his own engine lineage, and openly credits Claude as co-author. It differs from the expert, incrementally-developed Triumviratus (AI as a tool in a long human loop) by leaning further toward AI co-authorship: competent code, honestly disclosed, delivered as finished files.