Augustus

September 20, 2026 · View on GitHub

Place typed probabilistic judgment (Jev-class System One / decision models) using classical mental models. Jev is the exemplar, not the monopoly.

License: MIT Release Pages Claude Code Skills.sh

Homepage: 24601.github.io/Augustus

Agent skill for placing TypeSafe Jev Choice/Score/Noul with classical decision methods, composition algebra, and a validation gate.

Augustus, named for Augustus De Morgan (1806–1871), mentor and professor of William Stanley Jevons, is the design-judgment skill for where typed probabilistic judgment belongs (the Jev-class of System One models), using mathematical, logical, and algorithmic mental models. It applies across AI, software, business, knowledge work, and life, not only SWE. TypeSafe Jev is the documented exemplar (Choice, Score, Noul), not the monopoly. Formal methods are one pillar. Exact work stays in code or policy; the model owns narrow judgment; never launder a Noul as a proof.

Not a TypeSafe product. Companion, not replacement, to the official typesafe-ai skill. That skill owns Jev integration contracts; Augustus owns the design judgment: which pillar, family, and classical method map, what the objective implies for fail-open vs fail-closed, and what experiment would prove a design wrong. Not a TypeSafe-only how-to. Integrity / reward-hack companion: rh-guard.

Jev-class models with vs without Augustus. Without: call the model, act on the score, then quiet failure modes (soft Noul treated as hard gate, GPT bakeoff framing, no falsifier, polarity unchosen). With Augustus: state, pillar and family map, question design, fail-open vs fail-closed, typed Choice Score Noul, code owns effects, named falsifying experiment.

Jev-class: Jev, kev, Laya, OpenJev, GLiNER, SemIf, NanoJev, Jeff-1, localjev. Call and act, or place the judgment. Same split for any typed probabilistic judgment tool, not Jev-only.

Recipes

Class-wide, not a TypeSafe how-to. Full cards: docs/release-notes-v0.5.0.md · Pages recipes.

  • Decide vs generate. Without: treat tryDecide as another token stream. With: decide is not generate; typed calibrated judgments, not chat. Third-party benches stay theirs.
  • Encoder (GLiNER / GLiClass). Without: treat locate/categorize as a decision head and hard-gate spans. With: species map; remainder after extractive spans. Ports are class members, not Jev replicas. Measure span quality separately from ECE.
  • Open heads (Laya, SemIf, kev, Jeff-1). Without: wire-compat or argmax agree as replica. With: softmax ≠ calibrated Noul; systems timing ≠ semantic equivalence. Measure ECE/Brier on held-out, not only speed.
  • NanoJev. Without: game wins as calibration. With: specialist gameplay S1; local boolean ≠ TypeSafe noul. Measure held-out game separately from ECE.
  • llm-to-jev. Without: ship converted prompts as equivalent behavior. With: heuristic on-ramp; review the Score rubric. heuristic conversion ≠ calibrated Noul.
  • jcr. Without: run what the tree found. With: lookup returns context; does not execute. Routing ≠ permission; docs ≠ authority to run.
  • localjev / prompted JSON. Without: parse generated JSON as a Noul. With: schema-valid ≠ picked-right.

The skill

One line per file. The living catalog is in the reference cards and research/notes.md, not this README.

  • .agents/skills/augustus/SKILL.md: working protocol and decision-design card
  • .agents/skills/augustus/references/mental-models.md: cross-domain frames (EU, VOI, MCDA, SDT, ...); not SWE-only
  • .agents/skills/augustus/references/judgment-class.md: the class (Jev exemplar, not monopoly) and peer families
  • .agents/skills/augustus/references/formal-methods.md: judgment vs proof; soundness theater; DST trio
  • .agents/skills/augustus/references/formal-semi-formal.md: one-screen alias of the formal-methods pillar
  • .agents/skills/augustus/references/mixed-architecture.md: where S1 judgment sits next to LLM + code
  • .agents/skills/augustus/references/composition-algebra.md: positions a typed judgment can occupy relative to any method
  • .agents/skills/augustus/references/applied-mappings.md: sieves, keep/drop, triage, rank, and route placements
  • .agents/skills/augustus/references/faq.md: design-judgment FAQ (not an API how-to)
  • .agents/skills/augustus/references/mappings.md: classical-method mappings with boundaries and tests
  • .agents/skills/augustus/references/methods-catalog.md: named algorithms → judgment-shaped substitution
  • .agents/skills/augustus/references/toolbox-mapping.md: how to find a substitution in a method you already trust
  • .agents/skills/augustus/references/question-design.md: writing and diagnosing well-formed questions
  • .agents/skills/augustus/references/validation.md: design gate, eval recipes, Harbor/jevals practice
  • .agents/skills/augustus/references/boundary-audit.md: existing-system insertion: smallest boundary, red flags
  • .agents/skills/augustus/references/optimizer-integration.md: Jev inside Ax/DSPy optimizer loops
  • .agents/skills/augustus/references/agent-self-assessment.md: agent self-supervision gates (pre-action, done, stuck)
  • .agents/skills/augustus/scripts/evaluate_decisions.py: offline Brier / reliability / cost-threshold evaluator
  • research/notes.md: living hourly catalog (dense); §118 llm-to-jev conversion assistant (heuristic on-ramp, not a replica); §122 revisit / since-last-look protocol
  • research/README.md: evidence archive index (sources, refresh log, hourly dumps)
  • research/revisit-checklist.md: revisit already-catalogued repos when fingerprints move (revisit HIGH like novel HIGH)
  • research/changelog-hourly.md: hourly uniqueness dumps after v0.3.0
  • Hourly 0743 uniqueness dump: research/changelog-hourly.md (notes.md §113 / batch #96).
  • Hourly 0843 uniqueness lock: A hunch is a probability with a policy attached; { enter: 0.8, exit: 0.6 } is hysteresis; replay a policy change without inference; Decision models are providers, not the product; huncho ≠ Kungie/gut ≠ carldaws/hunch ≠ tpellet/hunch; pretrained Qwen2.5 base ECE 0.030 (0.5B) / 0.040 (7B); instruct 0.302 / 0.269; 70.9% → 70.0% mean conf 74.1% → 96.7%; temperature scaling still matches it in-distribution; No Jev API was called; Qwen2.5 ≠ Archer; Qwen/Qwen3.8-27B ≠ Archer; 学習済みモデル v0.1 は準備中です; bool AUROC 0.523; 先頭だと0件、末尾だと250件; 温度を渡さない場合、確率は較正されていません; このリポジトリには Jev を呼ぶコードが存在しません; g0runmezadam/what-is-jev IS tunahansahin897/what-is-jev (same GitHub id 1378007307); 947 repos scored; A 273 · B 302 · C 372; LLM rubric ≠ benches; Data as of 2026-09-20; HEAD 895b9498; README SHA 3ae98c56; 13 focused checks and one mutually exclusive outcome; Probabilities are advisory, not calibrated guarantees; omni-/ask-jev ≠ pedroknigge/mcp_jev; pd.cut bins by equal width while jeval bins by quantile; ECE 0.113 and ECE 0.076; jeval drift is not implemented yet; rlaope/jeval ≠ dayhaysoos/jevals; calibration does not compose; ECE has exactly zero statistical power to detect the failure mode that kills trajectories; 25–60× headline withdrawn; P(all-correct): 0.0071 vs 0.0001; TCE / AMS; Qwen 3.8 sparring ≠ Archer; Deferred Crispification; light_cutoff_applied_to_combination 0; recorded run, kinematic animation; BANKING77 Accuracy BERT-Base 93.02 Jev 79.90; Analyse jev calibration (NLL, ECE) backlog; BERT figures are published supervised references, not zero-shot; 档位措辞效应 分数极差中位 0.50、最大 1.32; 修好后对照组是 0.01; 不是 benchmark; 概率没做 calibration; ~1,430 API calls, about $0.15; xiaohuaxi/jev-study ≠ wjdjdakf17/jev-study ≠ baekenough/jev-study; AND: product (independence assumed and recorded in the trace); chat model's stated confidence is not calibrated; circuit-vl-4b ≠ Archer; Bring your own API key; vamsikrishna2421/jev-usecases ≠ whyashthakker/awesome-jev-use-cases; catalog ≠ endorsement; SemIf 2237★ (+30 vs §111 2207); jevlike 1049★ (+6 vs 1043); TypeLLM/TypeLLM 16★; AnotiaWang 98★; yibie/awesome-jev 520★ (+14 vs 506); Laya likes 861 (was 822); tracker likes 66 (+2 vs 64) lastModified UNCHANGED; Blackwood likes 2 gated manual; Archer still promised_not_landed; Hub archerhume/4rcherhume HTTP 401; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35; notes.md §114
  • User-provided 0915 uniqueness lock: TianyuCodings/NanoJev unified-games-v1 densify; A 0.6B parallel decision model: states and questions in, complete probability distributions out. Zero output-token decoding.; One model, four games; ViZDoom Basic 128/128 vs Jev 56/128; Predict Position 27/128 vs Jev 11/128; Maze 225 attempts vs Jev 2738; Snake 30 food / 256 steps; held-out Maze 4/10 Snake 8/8 Basic 128/128 Predict 27/128; Untuned Qwen3-0.6B baseline; 18,760 questions per variant; 16,333 ViZDoom; 896 Predict Position expert episodes; hard_lr1e5; mix weights 1/3, 1/3, 1/6, 1/6; Hub C-Tianyu/NanoJev revision unified-games-v1 likes 58; dataset C-Tianyu/NanoJev-Data likes 5; HEAD 618cea6d906d54e128360786d12f703fff2b1245; 1289★ / 158 forks / size 64035; README SHA 4190093c64ee75b26e9726daa3b00cbcf6d3157a; MIT; caijinchun/nanojev-arena ≠ liao96312/jev-arena-nanojev ≠ zwliJay/jev-forge ≠ NanoJev; not TypeSafe Jev; open replica / specialist gameplay S1; soft scores ≠ hard gates; Game success ≠ calibrated Noul; local type boolean ≠ TypeSafe noul; invented_signal false; do not reopen or amend PR #31/#32/#33/#35; notes.md §115
  • User-provided HIGH NiazMorshed2007/jcr (research/notes.md §116 / items 309–316 / batch #99). Capability-tree lookup; does not execute. Merged #35 owns §114. Merged #36 owns §115. Open #37 §117.

Install

Claude Code (plugin marketplace, mirrors the official TypeSafe layout):

claude plugin marketplace add 24601/Augustus
claude plugin install augustus@augustus

Any skills-compatible agent (Amp, Codex, Cursor, …):

npx skills add 24601/Augustus --skill augustus

Copy the skill path (Cursor / Amp / any agent that reads repo-local skills):

git clone https://github.com/24601/Augustus.git
# skill lives at .agents/skills/augustus/

ChatGPT: skills are not a native ChatGPT primitive. Paste .agents/skills/augustus/SKILL.md plus the references/ files into a GPT's instructions or a Project's knowledge and it will follow the protocol.

Amp: repo-local .agents/skills/ are discovered automatically.

GitHub topics

jev typesafe typesafe-ai system-one system-one-models structured-output calibrated-confidence ai-agents agent-skills decision-systems reranking beam-search claude-code python llm decision-theory decision-making semantic-search agent-workflows mixed-architecture agentic-ai zero-shot-classification tool-routing skill-routing semantic-lint classification gliclass listwise-ranking vision-scoring open-weights formal-methods model-checking deterministic-simulation value-of-information signal-detection mcda calibration alloy apalache pufferlib stamp-stpa

Versioning

See CHANGELOG.md and releases. Current: 0.5.0, written against typesafe-ai/skills v0.5.7 (65a39f3; live HEAD still this commit). Re-read live TypeSafe docs before treating that pin as current API behavior.

License

MIT. See LICENSE. Security reports: SECURITY.md. Contributions: CONTRIBUTING.md.

User-provided 0920 jcr uniqueness lock: NiazMorshed2007/jcr MIT; site https://jcr.niazmorshed.dev; topics ai-agents,jev,mcp; 4★; HEAD 138b3832; README SHA 2a49dbc1; LICENSE SHA 46231303; size 14850; Jev Capability Resolver; one tool to find documented deterministic commands in a nested capability tree; returns context; does not execute; skills = workflow+judgment; capabilities = individual operations; format independent of Jev; proposed open standard exploration; classify (Jev) → optional OpenAI decompose compound → beam search geometric mean of routing probs; keep up to 3 paths ≥60% of best (JCR_BAND_RATIO 0.6); ambiguity / no-match / depth-limit explicit; soft scores ≠ hard gates; 0.6 band is application policy; routing ≠ permission; docs ≠ authority to run; sol-vs-opus5-20 theirs: 20 scenarios × 4 variants = 80 runs; lookup+explain only, no execution; Claude Opus 5: agent input 108,585→15,819 (−85%), cost $0.3700→$0.1222 (−67%), wall 105.5s→77.7s; Codex GPT-5.6-Sol: 61,952→47,669 (−23%), $0.1377→$0.1151 (−16%), wall 25.3s→62.4s (Sol slower with JCR in 19/20); One Sol outlier 372.6s / 193 Jev calls; n=1 per cell; Not Harbor task-execution; Claude/Codex harnesses; compare mode; 50 scenarios bundled; 11 groups, 960 nodes, 11,360 items; 16 routing rounds per step; NiazMorshed2007/jcr ≠ skill-broker ≠ skillranker ≠ jev-sift ≠ jev-lens ≠ jevusher ≠ jev_select_capability; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34; notes.md §116 User-provided 0922 uniqueness lock: SemIf was formerly OpenJev; independent; not affiliated with Jev or TypeSafe; homepage openjev.com; default master; MIT; HEAD ca3ba65f1429; Tolerate float roundoff in MLX evidence verification; pushed 2026-09-19; live REST 2282★ / 140 forks; size 9177; README SHA 74ab7f7f; LICENSE SHA ca562883; interface pattern reproduction with open models; does not reproduce Jev undisclosed model/training; Direct option logits; 0 output tokens; shared-state parallel; MLX backend for Apple Silicon (--backend mlx); Qwen3.5-4B 3090 direct 1.023s vs AR JSON 5.332s (5.21×); argmax agree 18/21; systems comparison ≠ semantic equivalence; Parallel suffixes 20.03 dec/s on 777 decisions; Browser ladder Qwen3.5-4B authored BA 0.813, pert 0.766, TypeSafe subset agreement 0.845 vs Published Jev 0.883 (102 across 20 cases); Softmax over options ≠ calibrated Noul; typed output does not guarantee semantic correctness; wire/agreement ≠ replica of TypeSafe; SemIf ≠ kw2828/OpenJev playground ≠ zhihz/openjev ≠ apiplant/semif-rs port ≠ dddanielliu/semif-serve; rename is densify not a second census; JevBench 74.6 is §78 not this ladder; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35/#36/#38; do not push onto open #39/#40; notes.md §117 User-provided 0940 uniqueness lock: Turn decision-shaped LLM prompts into proposed Jev primitives; This is a conversion assistant, not an automatic guarantee of equivalent behavior; The compiler uses deterministic heuristics, not an LLM or evaluation model; It understands a deliberately small set of common prompt patterns; Generated instructions and criteria must be reviewed before production use; Score ranges such as 0 to 1 are translated into ordered Jev criteria; Prompts requiring open-ended prose are not a fit; suitability strong/partial/not_a_fit; compatibility full/partial/none; Writing new text stays with an LLM; Review the generated Score rubric; Jev scores ordered criteria, not an arbitrary 0-to-1 range; Everything runs locally in the browser; There is no framework, database, account, API, or server-side prompt processing; The key is read from the process environment and is never stored or printed; connect-src 'none'; alexwestco/llm-to-jev ≠ altryne/jevify ≠ ryana/jevify ≠ fidecastro/jevify ≠ Mintzs/jevify ≠ gulagala001/jevify ≠ uspraveen/Jevify; HEAD 234058ab372d; README SHA 43cd94fb; LICENSE SHA 5f334006; compiler SHA fdf235d0; 2★; MIT; JavaScript; size 29; Pages https://alexwestco.github.io/llm-to-jev/; invented_signal false; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35; notes.md §118 Hourly 0947 uniqueness lock: Fast and cheap agent evals. jev as judge.; 18,041 skills from the 200 most-starred repos; Not a security scanner; 最简 Jev 调用演示器; confidence 不是正确率; q93304989-bit/jev-lab ≠ tanayvasishtha/jev-lab ≠ dairui1/jev-lab ≠ BrendanH18/jev-lab ≠ yibie/laya-jev-lab; 75% cheaper and 18% faster withdrawn; jev @0.15 100% recall 87% savings; 33Audits/jev-auto ≠ gargpratyush/jev-router; no Typesafe key, no PI_API_BASE, zero deps; tool-emitted Score/Noul ≠ calibrated Noul; semantic_compatibility: false; candidate_mass; Qwen3.5-2B ≠ Archer; Jev evaluates decisions; it cannot run a coding-agent session; Status: no model yet; S1LV3RJ1NX/openjev ≠ TheoLeeCJ/openjev; 28 accepted decisions; 3 targets; score 800; health 100; arcade game not a flight trainer; A successful live TypeSafe call has not been verified for v0.1.0; abhibansal60/tidy ≠ MANISH007700/tidy; No model, Jev included, predicted which channels its owner keeps; seed 1 selected on a held-out 400-item validation split; Brier 0.342 → 0.378; more accurate and more overconfident; Qwen3.5-4B ≠ Archer; static quants of kushalpatil/jevify-gemma4-26b-a4b; The labels were corrected, and one earlier result was retracted; zero of 23,869 eligible rows; Do not compare cost without checking task success; Exit 1 is not a proof; kisshan13/typesafe-ai-go ≠ Nibir1/typesafe-go ≠ official; 38 tests that cannot fail in a 356-model warehouse; if a parser can answer it, Jev is never asked; 359 of them; Games & Simulation 82; Education & Learning 1; Ratings are heuristics; syedabbasshaheer-art/jev-atlas ≠ ZeroX-01/jev-atlas ≠ Zaious/jev-capability-atlas ≠ gorock007/jev-atlas; anandi1989/awesome-jev-usecases ≠ whyashthakker/awesome-jev-use-cases ≠ walidboulanouar/awesome-jev-use-cases ≠ vamsikrishna2421/jev-usecases; Every headline result above is self-reported; Archer Hume 84.6% MMLU-Pro is a third-party probe not landed Archer; catalog ≠ endorsement; judge ≠ actuator; softmax over A–H ≠ Noul; SemIf 2270★; jevlike 1054★; TypeLLM/TypeLLM 16★; AnotiaWang 98★; yibie/awesome-jev 538★; Laya likes 889; tracker likes 68 lastModified UNCHANGED; Blackwood likes 2 gated manual; Archer still promised_not_landed; Hub archerhume/4rcherhume HTTP 401; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#35/#36/#37/#38/#40; do not push onto open #39; notes.md §119

  • Hourly 1441 HIGH (research/notes.md §125 / items 417–432 / batch #107). openjev STE backends+Codiv / dual serving / jev-visual theirs / jev-mcp advisory. uniqueness_gate 0843+0915+jcr+0922+0940+0947+1049+1143+1248+1340+1441. Does not bump 0.5.0. Merged #47 owns §124. Merged #46 owns §123.
  • Hourly 1340 HIGH (research/notes.md §124 / items 401–416 / batch #106). typesafe-sdk 0.7 Pydantic / MLX 400 / PLAN_Qwen35 densify / GLiNER locate. uniqueness_gate 0843+0915+jcr+0922+0940+0947+1049+1143+1248+1340. Does not bump 0.5.0. Merged #46 owns §123. Merged #45 owns §122.
  • Hourly 1248 HIGH (research/notes.md §123 / items 385–400 / batch #105). decide is not generate / GLiNER-GLiClass ports / third-party benches theirs / openjev 0.2.1 densify. uniqueness_gate 0843+0915+jcr+0922+0940+0947+1049+1143+1248. Does not bump 0.5.0. Merged #45 owns §122. Merged #44 owns §121.
  • Hourly 1143 HIGH (research/notes.md §121 / items 369–384 / batch #104). open recreation / semantic lint sensor / third-party benches theirs / Laya serving densify. uniqueness_gate 0843+0915+jcr+0922+0940+0947+1049+1143. Does not bump 0.5.0. Merged #43 owns §120.
  • Hourly 1049 HIGH (research/notes.md §120 / items 353–368 / batch #103). ggmlc GGUF serving / option-order / exact-p Minesweeper / adapters / catalogs. uniqueness_gate 0843+0915+jcr+0922+0940+0947+1049. Does not bump 0.5.0. Merged #41 owns §119. Merged #42 is the 0.5.0 release.
  • Hourly 0947 HIGH (research/notes.md §119 / items 337–352 / batch #102). jev-as-judge / OneForward candidate_mass / catalogs / replay / RLCD heads. uniqueness_gate 0843+0915+jcr+0922+0940+0947. Does not bump 0.5.0. Merged #42 is the 0.5.0 release. Merged #40 owns §118. Merged #39 is §114 hygiene. Merged #37 owns §117. Merged #38 owns §116. Hourly 1049 uniqueness lock: ggmlc GGUF is not llama.cpp; Loading them in llama.cpp will fail; one encoder pass; hf:mys/laya-GGUF sha 713ae6f6e39f likes 0 apache-2.0; hf:mys/laya-multilingual-GGUF sha 3b645ae54281; hf:mys/laya-typed-decisions-GGUF sha 1e9e8ba1f527; hf:tozp/laya-onnx sha 0862aeba1e65 Opset 14 FP32 and INT8; tozp/laya-onnx ≠ Mattepiu/laya-onnx ≠ gqgs/laya-onnx; docker-laya MIT HEAD 1b8239a51ddd README SHA 9cb7bdc3; laya.cpp RTX ggml CUDA HEAD 8590937c79a2 README SHA cdd429b9; serving substrate ≠ calibrated replica; Softmax over options ≠ calibrated Noul; Qwen3.5-4B ≠ Archer; Qwen3.5-9B ≠ Archer; jev-position-test n=6 HEAD 7a56ca1c2698 README SHA 23f194c9; jevmlx slots 5 of 6; hosted Jev 0 of 6; prior_correction made it worse; jevSweeper mean Spearman ρ −0.274; picked exact-optimal 1/25 (4%); 31 of 36 still logically decidable; 86% of the time we should not have been asking; game success ≠ calibrated Noul; LLM2Jev 64★ Apache-2.0 HEAD 924618721277 README SHA da35fe61; not affiliated with or endorsed by Jev or TypeSafe; No answer tokens are generated; OpenSourceJev llama.cpp Qwen3-1.7B HEAD 3c41fba3681d; JEV-MLX Qwen3.5-9B HEAD dec24cd929ea; decision-head-rlcd Qwen3.5-4B 4.9M LoRA; AUTO_ACT is not a Noul; closed-set fail-open stdlib-only; verified=False; soft scores ≠ hard gates; 22 to 40% cheaper theirs; first version 70% more expensive; 111-case benchmark theirs; CDC 5,000 Jev 72.2% AUC 0.7725 theirs; accuracy is a trap; 9.0% base rate always-no 91.0%; catalog ≠ endorsement; jev-skill 109★ 90 scenarios HEAD 4f6e899a24d4; awesome-jev-live 673 entries 4★; minecraft-agent 214★ 131 JEV decisions 35 Astra calls; nether-final-08 8 minutes 43.300 seconds; planner writes JEV selects; RoboJEV structured simulator state not images; ashare-trader 策略未通过自己的回测门槛; 36 组参数全部净期望为负; no positive expectation under real costs; typed_evals NOT an official TypeSafe AI product; jev-as-judge is a sensor; third-person-audit 40% & 60% watermarks still soft; The included experience uses a handwritten demo provider; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35/#36/#37/#38/#39/#40/#41/#42; notes.md §120

Hourly 1143 HIGH (notes.md §121). open recreation ≠ calibrated replica. semantic lint is a sensor not a proof. cutoff 0.8 still soft. paired bootstrap CIs theirs. Same accuracy, 35x faster theirs. systems comparison ≠ semantic equivalence. serving substrate ≠ calibrated replica. catalog ≠ endorsement. permission ≠ confidence. Do not reopen or amend PR #23–#43. Does not bump 0.5.0. Skip Archer. invented_signal: false. Hourly 1143 uniqueness lock: open recreation ≠ calibrated replica; Qwen3.5-4B ≠ Archer; It is an open re-creation of Jev; less calibrated; perch 164★ MIT HEAD ba775a9940b6 README SHA 7ad0403b; semantic lint is a sensor not a proof; oxlint-plugin-jev cutoff 0.8 still soft; nassim-arifette/jevgrep ≠ Bentlybro/jevgrep ≠ can1357/jegrep ≠ uehaj/jev-semgrep; patdown fuzzy linter; PanAchy/jevvy ≠ Atominac/jevvy; No orders, no advice; SmartMoney-Cub 25★ HEAD d93cf493853d; paired bootstrap CIs theirs; emretheus/jev-rag-benchmark ≠ erendikmenn/jev-rag-benchmark; +0.82 pts XQuAD-EN 95% CI +0.35 to +1.31; +7.62 pts SciFact CI +4.88 to +10.38; Same accuracy, 35x faster theirs; systems comparison ≠ semantic equivalence; BANKING77 500 Jev 81.0% GPT-OSS 82.8% Mercury 73.2% Gemini 85.4% theirs; frozen cascade missed its evaluation accuracy target 430/500 vs GPT-5 432/500; This is not demonstrated equal-quality savings; 24 invented tickets; Routing errors caught by the gate 0 of 3; sample too small to establish calibration; This is not TypeSafe Jev; No real API requests were made; wire-compat ≠ replica; KonghaYao/laya-jev 按官方接口写的客户端只改一个 base URL; gqgs/laya-onnx densify 496.8 MiB; tozp/laya-onnx ≠ Mattepiu/laya-onnx ≠ gqgs/laya-onnx; serving substrate ≠ calibrated replica; BeatAPI/awesome-jev ≠ 99hansling/awesome-jev ≠ Vishnurr2k01/awesome-jev ≠ robokrunch/awesome-jev ≠ rudy2steiner/awesome-jev-hub; All 125 projects; catalog ≠ endorsement; Pasblinn/jev-lab ≠ tanayvasishtha/jev-lab ≠ dairui1/jev-lab ≠ BrendanH18/jev-lab ≠ yibie/laya-jev-lab ≠ q93304989-bit/jev-lab; Independent project. Not affiliated with TypeSafe; Kevthetech143/super-jev densify experimental V0.2.0; permission ≠ confidence; allay-team/openjev ≠ piyush-infocusp/openjev ≠ TheoLeeCJ/openjev; 2022 Mineflayer Jevalent collision; kushalpatil/jevify-gemma4-e4b GGUF densify; static quants; This dataset and model are independent research artifacts, not reproductions of Jev or RLCD; pngwn demo accuracy 0.705 ECE 0.046 ~112 ms theirs; cutoff 0.8 still soft; soft scores ≠ hard gates; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35/#36/#37/#38/#39/#40/#41/#42/#43; notes.md §121

Hourly 1248 HIGH (notes.md §123). decide is not generate. tryDecide returns typed calibrated judgments not a token stream. GLiNER/GLiClass ports are class members not Jev replicas. 93.5% theirs not Harbor. 74.9 theirs not Harbor. 8.7x theirs not Harbor. wire-compat ≠ logit-equiv. SHA move is not a replica. catalog ≠ endorsement. Do not reopen or amend PR #23–#45. Does not bump 0.5.0. Skip Archer. invented_signal: false. Hourly 1248 uniqueness lock: decide is not generate; tryDecide returns typed calibrated judgments not a token stream; juspay/neurolink 133★ MIT HEAD 268b0fe83130 README SHA e709cadfa6b6 tag v12.19.0; GLiNER/GLiClass ports are class members not Jev replicas; MacPaw/Gliner2Swift ≠ Knowledgator/GLiClass.c ≠ fbilhaut/gliclass-rs ≠ Knowledgator/GLiClass.js ≠ gravitee-io/GLiNER4j ≠ apiplant/gliner-rs ≠ codesoda/gliner2-rs; 8.7x faster 4.4x fewer prompts theirs; 153 was a reporting error; corrected 156-case 9.8x faster 4.2x fewer prompts theirs; independent v0.2.1 1.24x vs Mini theirs; Approvals only; anpicasso/hermes-jev-approvals ≠ hermes-switchyard; scx-router GLiClass ranks candidate LLMs in one non-generative pass; typesafeai-dotnet-sdk Not affiliated with TypeSafe AI; hyperspaceai/jevcache ≠ kushals256/jevcache; ST-jeved measures each reply; 400 plain-text for unaskable question; razorback16/openjev:0.2.1 Docker densify HEAD 794a81b87131; wire-compat ≠ logit-equiv; Option-Marker joint attention 93.5% macro theirs; 93.6% micro theirs; n=78; T = 1.0367 vs T = 1.1692 two temperatures; guaranteeing is soundness theater; wfzyx/von densify HEAD bed7e7337791; Benchmark Heaven leaderboard #2 74.9 theirs; NLL calibration assets; 77.10% still §71 claim-audit; do not re-fold as a beat; Heman10x-NGU/openJev-verdict-2.0 densify HEAD bff28567cff4; kev-family weight tarballs; PLAN_Qwen35 proposal for review; deadline 0.53→0.82 at 9B theirs; Qwen3.5-9B ≠ Archer; isolation would fail by construction on DeltaNet; jaredpalmer/kev densify; JevBench v1.2.2 jeff 66.9 (#9) jev 75.3 (#2) theirs; logan-markewich/jeff densify HEAD 34b32f99a727; thinking=True/False per-field budget; type safety does not guarantee factual accuracy; TypeLLM/TypeLLM densify HEAD c4b03ba9e792; us/jev-local stub until hf; Eran-BA/Jev_from_GLiNER2 spec ≠ replica; lsu-ub-uu/systemone ≠ TypeSafe System One; Layan/Laya HF spaces name-match; catalog ≠ endorsement; decide ≠ generate ≠ stream; 93.5% theirs not Harbor; 74.9 theirs not Harbor; 8.7x theirs not Harbor; SHA move is not a replica; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35/#36/#37/#38/#39/#40/#41/#42/#43/#44/#45; notes.md §123

Hourly 1340 HIGH (notes.md §124). typesafe-sdk 0.7 Pydantic response models. msgspec dropped. MLX backend 400 plain-text error contract. Pydantic response models ≠ logit-equiv. msgspec dropped is not a replica. Error contract is not a Noul. PLAN_Qwen35 densify. coverage-at-error-budget theirs not Harbor. GLiNER locate ports are class members not Jev replicas. Locate ≠ decide. ~160 ms theirs not Harbor. SHA move is not a replica. catalog ≠ endorsement. Do not reopen or amend PR #23–#46. Does not bump 0.5.0. Skip Archer. invented_signal: false. Hourly 1340 uniqueness lock: typesafe-sdk 0.7 Pydantic response models; msgspec dropped; The server's output is unchanged and was never wrong; MLX backend 400 plain-text error contract; SchemaError is 400 plain-string detail not 422 list; razorback16/openjev densify HEAD 6e91dfc031bc README SHA cbdcc8de0304; Pydantic response models ≠ logit-equiv; msgspec dropped is not a replica; Error contract is not a Noul; wire-compat ≠ logit-equiv; PLAN_Qwen35 densify; corrected Qwen3.5 LoRA target names verified; in_proj_qkv in_proj_z in_proj_a in_proj_b out_proj; peft 0.21 existence proof; OOD-calibration study; coverage-at-error-budget metric in Phase 0; PLAN_Qwen35 still proposal for review; deadline 0.53→0.82 at 9B theirs; isolation would fail by construction on DeltaNet; Qwen3.5-9B ≠ Archer; jaredpalmer/kev densify HEAD 75cc15ddb8e2 PLAN SHA eca543246f50; GLiNER locate ports are class members not Jev replicas; urchade/GLiNER ≠ fbilhaut/gline-rs ≠ lmoe/gliner-onnx.js ≠ shershah1024/gliner-native-runtime; Locate ≠ decide; Jev-Vision skip 0.936 effect 0.967 done 0.896 157 ms theirs; ~160 ms theirs not Harbor; 0.971 F1 theirs not Harbor; coverage-at-error-budget theirs not Harbor; hf:fr0stbit3/laya-gguf serving substrate ≠ calibrated replica; jkcdarunday/SystemOne-Next ≠ TypeSafe System One; catalog ≠ endorsement; SHA move is not a replica; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35/#36/#37/#38/#39/#40/#41/#42/#43/#44/#45/#46; notes.md §124

Hourly 1441 HIGH (notes.md §125). vLLM NVIDIA + MLX Apple Silicon. Codiv hosted free endpoint. dual /v1/systemone + /v1/chat/completions. chat 501 on MLX. dual serving is not generate. Hosted Codiv ≠ TypeSafe. candidate probabilities are relative not correctness. recommendation is advisory. the server never blocks on its own. LoRA ≠ RLCD replica. pass-min 0.8 still soft. 37.30s → 2.40s at 64 decisions theirs. 2B 94.71% 9B 97.54% hard test theirs. wire-compat ≠ logit-equiv. SHA move is not a replica. catalog ≠ endorsement. Do not reopen or amend PR #23–#47. Does not bump 0.5.0. Skip Archer. invented_signal: false. Hourly 1441 uniqueness lock: vLLM NVIDIA + MLX Apple Silicon; Codiv hosted free endpoint; dual /v1/systemone + /v1/chat/completions; razorback16/openjev densify HEAD cddbd962c88a README SHA a5943415cb92; STE README rewrite; serving-port densify; chat 501 on MLX; dual serving is not generate; Hosted Codiv ≠ TypeSafe; wire-compat ≠ logit-equiv; SHA move is not a replica; Error contract is not a Noul; hr98w/jev-visual 167★ Apple Silicon visual candidate scoring; 37.30s → 2.40s at 64 decisions theirs; Breakout 9 bricks 6 returns 2 lives theirs; candidate probabilities are relative not correctness; jkudish/jev-mcp 156★ ten MCP tools; recommendation is advisory; the server never blocks on its own; TypeSafe CLERC 5% to 18% theirs; jkudish/jev-mcp ≠ burnigtm/jev-mcp; zhengxuyu/litjev off-the-shelf Qwen decision layer; Probabilities are not calibrated by default; Qwen/Qwen3.8-27B ≠ Archer; zhengxuyu/litjev ≠ alexwestco/llm-to-jev; Zefan-Cai/Open-Jev LoRA + scalar head; 2B 94.71% 9B 97.54% hard test theirs; 2B OOD 86.02% 9B OOD 91.97% theirs; 80,816 training rows; 27B still in progress; LoRA ≠ RLCD replica; Zefan-Cai/Open-Jev ≠ TheoLeeCJ/openjev ≠ razorback16/openjev; cristianoliveira/jeq intelligence you can pipe; pass-min 0.8 still soft; JEQ does not own actions; AndyInQtr/laya-coreai CoreML serving substrate ≠ calibrated replica; AndyInQtr/laya-coreai ≠ mizorewww/laya-coreml; catalog ≠ endorsement; Archer still promised_not_landed; Hub archerhume/4rcherhume HTTP 401; do not reopen or amend PR #23/#24/#25/#26/#27/#28/#29/#30/#31/#32/#33/#34/#35/#36/#37/#38/#39/#40/#41/#42/#43/#44/#45/#46/#47; notes.md §125