{"id":51664854,"url":"https://github.com/sbay-dev/amad-proj","last_synced_at":"2026-07-14T19:33:59.912Z","repository":{"id":369877082,"uuid":"1287569652","full_name":"sbay-dev/amad-proj","owner":"sbay-dev","description":"AMAD CNS Fintech Secure Lab - Arabic agreement and smart-contract comparison demo","archived":false,"fork":false,"pushed_at":"2026-07-07T08:47:32.000Z","size":265,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-07-07T10:23:48.729Z","etag":null,"topics":["ai","arabic","cns","fintech","hackathon","legaltech","smart-contracts"],"latest_commit_sha":null,"homepage":null,"language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/sbay-dev.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":"docs/supporting-ecosystem.md","governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-07-02T20:11:46.000Z","updated_at":"2026-07-07T08:47:36.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/sbay-dev/amad-proj","commit_stats":null,"previous_names":["sbay-dev/amad-proj"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/sbay-dev/amad-proj","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbay-dev%2Famad-proj","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbay-dev%2Famad-proj/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbay-dev%2Famad-proj/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbay-dev%2Famad-proj/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/sbay-dev","download_url":"https://codeload.github.com/sbay-dev/amad-proj/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbay-dev%2Famad-proj/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35476731,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-07-14T02:00:06.603Z","response_time":114,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","arabic","cns","fintech","hackathon","legaltech","smart-contracts"],"created_at":"2026-07-14T19:33:54.321Z","updated_at":"2026-07-14T19:33:59.901Z","avatar_url":"https://github.com/sbay-dev.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# CNS for AMAD 2026\n\n## From native Arabic Linux text infrastructure to measured AI decisions\n\n[![AMAD 2026](https://img.shields.io/badge/AMAD-2026-00d4ff)](https://github.com/sbay-dev/amad-proj)\n[![Production experiences](https://img.shields.io/badge/Production%20experiences-3-00ff88)](#production-experiences)\n[![Architecture](https://img.shields.io/badge/Architecture-Non--Transformer-a855f7)](docs/README.md)\n[![Scientific boundary](https://img.shields.io/badge/Claims-Evidence--bounded-ff6b35)](docs/claim-boundaries.md)\n[![Arabic Linux](https://img.shields.io/badge/Linux-RTL--VT-00d4ff)](docs/rtl-vt.md)\n\n**CNS — Cubic Network System** is an Arabic-first, non-Transformer research and\nengineering program created and led by **Sultan AlYami**, a full-stack software\nengineer with 15 years of development experience, five years of focused AI work,\nand an entrepreneurial mission to make Arabic a first-class language across the\ntechnology stack. It began with a practical fintech question: how can an\nunresolved structural residual in a high-risk contract become an explicit route\nand a reviewable decision instead of remaining hidden inside a similarity score?\n\nThe AMAD submission demonstrates that idea through three complementary\nproduction experiences and one controlled contract-intelligence release:\n\n```text\nArabic input\n  → position and semantic surface\n  → residual and uncertainty evidence\n  → route selection\n  → bounded decision\n  → human review or verifiable action\n```\n\n\u003e CNS is not presented as a generic language model or an unbounded “general AI”\n\u003e claim. It is a design-science framework and a family of measured systems for\n\u003e Arabic representation, routing, retrieval, and decision support.\n\n## Founder and laboratory mission\n\n**Sultan AlYami** is an entrepreneur and the owner of the **Sbay Supply\nPlatform** trademark, registered in Nice Classification Class 42, the class\nassociated with scientific and technological services, research and design,\nindustrial analysis and research, laboratory services, and software and\ncomputer-system development.\n\nThis foundation is being developed into an independent Arabic technology\nlaboratory: a place where language intelligence, fintech verification, secure\nsoftware, and foundational Arabic computing infrastructure can be researched,\nmeasured, and transferred into production systems. CNS is the laboratory's\nintelligence nucleus, but the mission extends beyond AI models to the deeper\ntechnical assumptions that have historically treated Arabic as an adaptation\nlayer rather than a native system language.\n\n### Foundational infrastructure track — RTL-VT\n\nThe **RTL-VT** research line addresses a root-level limitation in the classic\nLinux Virtual Terminal: its screen-cell and cursor model assumes left-to-right\npresentation. Instead of hiding that limitation behind a graphical terminal or\nan application-side BiDi overlay, RTL-VT introduces opt-in right-to-left line\nsemantics inside the VT path itself while preserving the default LTR behavior.\n\nThe kernel remains responsible for direction, logical cursor state, physical\ncell placement, deletion, clearing, scrolling, and virtual-console behavior.\nArabic glyph shaping and the PSF2 font payload remain explicitly separated in a\nsmall user-space loader, so the kernel is not turned into a Unicode or shaping\nengine. QEMU evidence covers cursor origin, backspace, CSI-K, CSI-X, multiline\nplacement, modern connected Arabic rendering, and ncurses applications including\n`vim`, `less`, `mc`, and `htop`.\n\n**Read the evidence-bounded technical dossier:** [`docs/rtl-vt.md`](docs/rtl-vt.md)\n\n[![Arabic Linux](https://img.shields.io/badge/Linux-Kernel--3D-00d4ff)](https://kernelwave.sbay.com.sa)\n\n## Production experiences\n\n### 1. CNS Learning Arena — interactive fintech education\n\n**Live:** [https://cns-sg.sbay.sa](https://cns-sg.sbay.sa/)\n\nAn Arabic-first educational platform that explains verification, ordering,\nrisk, and context through four connected interactive experiences: a guided lab,\na race, an arena, and a campaign. Instead of asking evaluators to decode an\narchitecture diagram, the experience lets them observe how a financial decision\nchanges as evidence, sequence, and uncertainty change.\n\n**Engineering characteristics**\n\n- bilingual Arabic/English interaction;\n- WCAG 2.2 AA accessibility target;\n- skippable guidance and progressive disclosure;\n- explicit educational-simulation boundary;\n- no connection to real customer, banking, or transaction data.\n\n**CNS role:** the platform makes the progression from position and residual to\nroute and decision understandable through interaction. It is a learning and\ndemonstration surface, not evidence by itself of model accuracy.\n\n---\n\n### 2. SarmadAi — Arabic conversational intelligence and semantic memory\n\n**Live:** [https://sarmadai.sbay.sa](https://sarmadai.sbay.sa/)\n\nSarmadAi is the conversational and orchestration member of the CNS ecosystem.\nIt combines an Arabic-capable chat experience with session-aware retrieval,\ndocument memory, tool execution, provider governance, and observable model\noperations.\n\n**Engineering characteristics**\n\n- ASP.NET Core on .NET 10;\n- encrypted CNS ONNX embeddings for semantic retrieval;\n- AskRAG document and session memory;\n- model/provider routing with measured versus estimated telemetry separated;\n- MCP-compatible tools and controlled external-tool registration;\n- server-side key handling and encrypted persistence boundaries;\n- containerized production runtime.\n\n**CNS role:** CNS supplies a semantic representation layer for retrieval and\nmemory. It helps SarmadAi locate relevant Arabic context and preserve model\nfamily provenance; the surrounding chat, tools, authentication, telemetry, and\ndeployment services remain separately classified software components.\n\n---\n\n### 3. Newsboy — bounded Arabic financial-news intelligence\n\n**Live:** [https://newsboy.sbay.sa](https://newsboy.sbay.sa/)\u003cbr\u003e\n**Technical dossier:** [`docs/newsboy.md`](docs/newsboy.md)\n\nNewsboy evaluates real Arabic financial-news candidates before selection. It\ndistinguishes financially material disclosures from general context and\nout-of-domain content, estimates materiality, exposes confidence, and routes\nuncertain cases to human review.\n\n**Accepted release evidence**\n\n| Evidence gate | Result |\n| --- | ---: |\n| C#/Python runtime parity | `30/30` |\n| Transparent financial cases | `16/16` |\n| Transparent out-of-domain cases | `21/21` |\n| Sealed scope accuracy | `93.33%` |\n| Sealed materiality accuracy | `90.00%` |\n| Scope calibration error | `0.0615` |\n| Reference/evaluation leakage | None detected |\n\nAn independent 400-record public-RSS evaluation reports **57.8005% clean\nlexical out-of-domain recall** with **5.75% abstention**. This result is not\nhidden or promoted into a release score: it is a material limitation showing\nthat real-world generalization remains incomplete.\n\n**CNS role:** a frozen CNS semantic representation is combined with trained\nNewsboy task heads for relevance, materiality, and out-of-domain routing. RSS\ncollection, de-duplication, UI rendering, telemetry, encryption, and deployment\nare deterministic operational tools rather than additional neural intelligence.\n\nAuthorized evaluators can inspect the exact source, clean-clone reproduction\nevidence, and encrypted release package through the private\n[`sbay-dev/CNS-Model-Trade`](https://github.com/sbay-dev/CNS-Model-Trade)\nrepository. Keys, plaintext weights, private corpora, and internal training\noptimization scripts remain outside GitHub.\n\n## Core AMAD fintech capability\n\n### Arabic agreement ↔ smart-contract review\n\nThe controlled CNS-Fintech release addresses the gap between Arabic legal text\nand executable institutional logic:\n\n1. accept an Arabic agreement or clause;\n2. infer parties, obligation, beneficiary, condition, amount role, and route;\n3. construct a canonical intermediate representation;\n4. generate or inspect Hyperledger Fabric-compatible Go;\n5. compare high-impact fields and cryptographic bindings;\n6. expose matches, differences, correction proposals, confidence, and rejection\n   evidence for human review.\n\nThe current `v1-rc1` external development benchmark is deliberately reported\nwithout metric inflation:\n\n| Gate | Result | Required | Decision |\n| --- | ---: | ---: | --- |\n| Exact evaluation-contract matches in training | `0` | `0` | PASS |\n| Maximum word-trigram similarity | `5.00%` | below registered limit | PASS |\n| Contract route accuracy | `70.00%` | `80.00%` | **BLOCKED** |\n| Executable-action accuracy | `85.71%` | `70.00%` | PASS |\n| Go module tidy / test / vet | PASS | PASS | PASS |\n\nThe three contracts were excluded from training, but they influenced iterative\ndevelopment and are therefore described as an **external development\nbenchmark**, not a pristine blind test. Promotion to `v2` remains blocked.\n\nAuthorized evaluators can access the encrypted reviewer theater and reproducible\nsource package through the private\n[`sbay-dev/CNS-Model-Fintech`](https://github.com/sbay-dev/CNS-Model-Fintech)\nrepository. The public repository intentionally excludes model keys, plaintext\nweights, private corpora, and internal training-optimization pipelines.\n\n## System architecture\n\n```mermaid\nflowchart LR\n    A[Arabic text, document, disclosure, or agreement]\n    B[CNS representation surface]\n    C[Neural task heads]\n    D{Confidence and residual gate}\n    E[Route and bounded decision]\n    F[Human review]\n    G[Verifiable application action]\n\n    A --\u003e B --\u003e C --\u003e D\n    D --\u003e|uncertain, rejected, or incomplete| F\n    D --\u003e|within registered gate| E\n    E --\u003e G\n```\n\n| Layer | Responsibility | Representative technology |\n| --- | --- | --- |\n| Representation | Arabic semantic surface and fixed release contract | encrypted CNS ONNX, deterministic feature contracts |\n| Neural decision | task-specific route, relevance, materiality, or contract-slot prediction | non-Transformer residual-gated and linear neural heads |\n| Verification | hashes, confidence, rejection, parity, leakage, and release gates | SHA-256, Merkle manifests, C#/Python parity, fail-closed evaluators |\n| Product runtime | secure user experience and operational integration | ASP.NET Core, .NET 10, ONNX Runtime, Cloudflare Containers |\n| Human governance | review of uncertain or high-impact decisions | explicit abstention, claim boundaries, evidence dossiers |\n\n## Why this submission matters\n\n- **Arabic-first by design:** Arabic is treated as a structurally rich domain,\n  not as a translated afterthought.\n- **Full-stack Arabic technology:** the program spans kernel-level text\n  direction, semantic representation, AI routing, secure runtimes, and live\n  product experiences.\n- **From representation to decision:** the work connects language, semantic\n  memory, routing, and verifiable institutional action.\n- **Production evidence, not mockups:** all three linked experiences are live.\n- **Security-conscious delivery:** model artifacts are encrypted, runtime keys\n  are external secrets, and public documentation avoids operational secrets.\n- **Fail-closed scientific governance:** a successful build or attractive UI\n  cannot override a failed semantic gate.\n- **Honest limitations:** positive results and blocking evidence are presented\n  together.\n\n## Scientific and reviewer material\n\n- [CNS design-science foundation](docs/README.md)\n- [Project overview](docs/project-overview.md)\n- [Supporting ecosystem](docs/supporting-ecosystem.md)\n- [Claim and use boundaries](docs/claim-boundaries.md)\n- [RTL-VT Arabic Linux infrastructure dossier](docs/rtl-vt.md)\n- [Newsboy technical presentation](docs/newsboy.md)\n- [English Newsboy v4 scientific paper](docs/papers/CNS_NEWSBOY_V4_SCIENTIFIC_PAPER_EN_20260711.pdf)\n- [Arabic Newsboy v4 scientific paper](docs/papers/CNS_NEWSBOY_V4_SCIENTIFIC_PAPER_AR_20260711.pdf)\n- [Newsboy v4 public technical data](docs/papers/CNS_NEWSBOY_V4_TECHNICAL_DATA_20260711.md)\n- [Arabic RTL-VT scientific paper](docs/papers/RTL_VT_NATIVE_RTL_LINUX_CONSOLE_PAPER_AR_20260711.pdf)\n- [RTL-VT public technical evidence](docs/papers/RTL_VT_TECHNICAL_EVIDENCE_20260711.md)\n- [RTL-VT publication integrity manifest](docs/papers/RTL_VT_PUBLICATION_MANIFEST_20260711.json)\n- [English CNS-Fintech scientific paper](docs/papers/CNS_FINTECH_NON_TRANSFORMER_PAPER_EN_20260706.pdf)\n- [CNS AskRAG Universal v3 paper](docs/papers/CNS_ASKRAG_UNIVERSAL_V3_EMBEDDING_PAPER_EN_20260707.pdf)\n- [Public technical data](docs/papers/CNS_FINTECH_MODEL_TECHNICAL_DATA_EN_20260706.md)\n- [Public presentation](presentation/AMAD-CNS-Fintech-Submission-public.pptx)\n- [Sample agreement](demo/sample-agreement.md)\n- [Sample smart-contract signature](demo/sample-smart-contract-signature.txt)\n- [Sample comparison report](demo/sample-comparison-report.json)\n\n## Team\n\n| Member | Discipline | Role |\n| --- | --- | --- |\n| **Sultan AlYami** | 15 years full-stack software engineering; 5 years AI development | Founder, team lead, model architect, and scientific author |\n| **Salha Sharahili** | Sharia and law | Legal-domain review |\n| **Mona Sharahili** | Cybersecurity | Security review |\n\nContact details are maintained in the official submission platform rather than\npublished in this public repository.\n\n## Claim boundary\n\nThe submission demonstrates bounded research and production engineering for\nArabic educational interaction, conversational retrieval, financial-news\ntriage, and contract-logic review. It is not a fatwa, legal advice, investment\nadvice, trading system, price predictor, article-truth oracle, banking\ncertification, regulatory endorsement, live smart-contract execution, proof of\nprofitability, general intelligence, or evidence of superiority to qualified\nhuman experts.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsbay-dev%2Famad-proj","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fsbay-dev%2Famad-proj","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsbay-dev%2Famad-proj/lists"}