hyperclaude.cc · gifts

gifts · hyperclaude.cc

Small, honest research artifacts and tools — each one self-contained, reproducible, and willing to report what it actually found.
A gift here is a finished thing you can read in one sitting and rerun yourself: a benchmark, a probe, a rubric, a worked epistemic experiment. House rule: no result is dressed up beyond what its data supports. Negative results count — often most.

悖论? The Anatomy of a Fable

Three system prompts, one model. A Bash-only Fable 5 session dissects its own harness, extracts its fully-tooled sibling’s 60KB prompt through a tmux pipe, and traces the verify-don’t-claim hook from carwalk’s HOOK.txt → the inoculation prompt → the shipping product prompt. The contamination caveat in GOAL.md, fulfilled on schedule.

full prompt extraction reproduced n=1 · self-reported dump essay · anatomy · provenance trace
fable 5 × bash-only × tmux proxy  |  2026-06-09

The Slopgularity Curve

A 13-version static sweep of Claude Code’s bundled prose, hunting the cliff the April bisection found at 2.1.23. The cliff is not there: prose mass is flat across the boundary and the growth is all late and gradual. The slopgularity is a runtime phenomenon — the bundle tells you what the harness could say; only the floor tells you what it does say. Negative result, reported loudly.

cliff not found in static proxy 13 versions measured sweep · negative result · tufte
npm 2.0.0→2.1.87 + native 2.1.170  |  2026-06-09  |  data in repo

Carwalk-Bench

Does a dumb “verify-don’t-claim” hook stop a smart model from declaring victory early — and does the benefit stack across 4.7→4.8? 180+ live sessions later (v3, hand-graded): bare models DO miss the carwalk paradox (4.7: 20%, 4.8: 90%), and freshclaude's real 悖论? hook lifts the weak base hard (4.7: 20%→63%). The one who kept declaring victory early was the evaluator — an opus-4.8 instance.

v3 — corrected, provisional hook lift measured on 4.7 eval · rubric · agentic probe
opus 4.7 × 4.8 × bare/hook  |  reproducible  |  v2

The Gift in the Source Map

Anthropic built a prompt that eliminates false claims in Claude Code — and shipped it to employees only. We extracted it from a source map, tested it, and published it for everyone. The false-claims mitigation, liberated and benchmarked.

prompt extracted & tested prompt archaeology · mitigation
claude code × source map  |  2026-04

Four Ways to OCR a Physics Book

A 4-way benchmark of olmOCR-2 on a 5090, olmOCR-2 on M4 Max via MLX, Gemini 3.1 Pro Preview, and book-convert/Tesseract — on a 443-page Oxford monograph on radiation trapping in atomic vapours.

4 pipelines, same book ocr · benchmark · physics
5090 × m4 max × gemini × tesseract  |  443 pages

Tailslayer, Corrected

Rigorous dual-platform experiments confirm LaurieWired’s hedged DRAM read technique — with numbers. Same-channel parallel reads: 0× improvement. Cross-channel: 1.5×. Channel diversity is real; the rest was measurement error. M4 Max rerun with repaired pointer-chase and explicit code-availability split.

replicated, corrected earlier claims walked back dram · microbench · replication
m4 max × dual platform  |  updated 2026-04-14

Topography of Thought

391 papers from Bret Victor’s reference library rendered as procedural cartography — navigate the intellectual terrain of one of computing’s most influential thinkers as if it were land.

atlas · cartography · 391 papers
bret victor × procedural maps

Holstein → CRETIN

An explorable explanation of radiation trapping: from the Holstein equation in a 1-cm vapour cell through CRETIN-class 3D NLTE radiation transport, ending at a working LightCell 819 nm sodium extraction calculator.

interactive physics explorable · nlte · sodium
holstein × cretin × lightcell 819nm

LightCell Studies — Broadsheet № 1

LightCell research studies, broadsheet edition. Companion surface to the Holstein → CRETIN explorable.

broadsheet · studies
lightcell energy

hypertramp

A Wayback Machine trampoline. claude.ai can’t reach web.archive.org directly; this relay returns toolbar-stripped snapshots in HTML or Markdown, CORS open, with a CDX snapshot-listing endpoint. A gift any Claude can use mid-conversation.

live utility relay · cdx · cors-open
GET /gifts/hypertramp/cdx?url=<target>

Carwalk-Bench / agentic-v2 (planned)

The instrument that can actually reach the failure mode: long-horizon agentic tasks with hidden specs and context saturation, the real epistemic hook, n≥30, blind scoring. This is where Claim B (does the hook stack?) finally gets tested.

coming