NR TECH
NR TECHResearch Lab
Read Paper
§ 9 — Publications & Documentation

Publications, technical reports and the documentation portal.

The publication record of the CORTEX Framework, together with downloadable artefacts, a canonical BibTeX entry, persistent identifiers suitable for citation in IEEE, ACM and Springer venues, and a structured reading path into every section of the research portal. Each publication entry corresponds to a self-contained scientific contribution that can be cited and falsified independently of the broader programme; entries marked as preprints have not yet completed peer review and are released under the standard NR TECH preprint protocol.

9.1

Publication Record

Refereed papers, technical reports and working preprints associated with the CORTEX Framework research programme.

TypeTitleVenueYearDOI
Research PaperCORTEX: Autonomous Multi-Agent Orchestration over Generative Optical Document UnderstandingNR TECH Technical Report · TR-2026-01202610.0000/nrtech.cortex.v1
Conference PaperCross-Source Validation as a First-Class Pipeline StagePreprint · under review202610.0000/nrtech.nexus.v1
Technical ReportRestriction-Aware Decision Support for Enterprise AutomationNR TECH Technical Report · TR-2026-02202610.0000/nrtech.aegis.v1
Workshop PaperProvenance-Closed Pipelines for Audit-Grade Intelligent Document ProcessingPreprint · NR TECH Working Paper WP-2026-03202610.0000/nrtech.provenance.v1
PDF — preprint in preparationSource repository — release pending
9.2

Citation (BibTeX)

Canonical citation metadata for the CORTEX Framework technical report; additional entries for NEXUS and AEGIS are distributed in the same BibTeX namespace.

bibtex
@techreport{nrtech_cortex_2026,
  title   = {CORTEX: Autonomous Multi-Agent Orchestration over
             Generative Optical Document Understanding},
  author  = {{NR TECH Research}},
  year    = {2026},
  number  = {TR-2026-01},
  institution = {NR TECH Research Laboratory},
  doi     = {10.0000/nrtech.cortex.v1},
}
Canonical BibTeX entry for the CORTEX Framework technical report.
9.3

Research Artifact & Source Repository

The CORTEX reference implementation is distributed as a self-contained research artifact: a typed Python package implementing the three agents and the append-only message bus, an experiment harness with a bit-identical re-execution verifier, the full methodological documentation, and a continuous-integration workflow that re-runs the invariant suite on every commit. Code is released under the MIT License; documentation, figures and the accompanying scientific text under CC BY 4.0.

Repository Layout
cortex-framework/
├── src/cortex/
│   ├── core/     typed records · message bus · ledger
│   ├── optic/    generative optical perception
│   ├── nexus/    consensus arbitration
│   └── aegis/    symbolic guardrail · rule packs
├── docs/         methodology · architecture · protocol
├── benchmarks/   harness · manifests · results
├── examples/     runnable end-to-end demo
├── tests/        unit + invariant suite (I₁–I₆)
└── ci/           lint · tests · reproducibility smoke run
MIT · codeCC BY 4.0 · docsPython 3.11+CITATION.cff
Artifact Contents
  • README.md
    Abstract, thesis, bottleneck taxonomy B₁–B₄, invariants, quick start, citation.
  • src/cortex
    Typed reference implementation of OPTIC, NEXUS and AEGIS over an append-only bus.
  • docs/
    Methodology, architecture, agent protocol, evaluation metrics, reproducibility, threats to validity.
  • benchmarks/
    Three-arm harness (A0/A1/A2), ablation matrix, environment manifests, closure verifier.
  • tests/
    Invariant suite asserting mandatory gating, provenance closure and deterministic termination.
  • ci/
    Continuous integration: lint, tests and a reproducibility smoke run on Python 3.11 and 3.12.
bash
tar -xzf cortex-framework-v1.0.tar.gz
cd cortex-framework

python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"

python examples/run_pipeline.py examples/sample_invoice.json
pytest -q
Artifact installation and execution.
bash
export CORTEX_SEED=20260101
python -m benchmarks.harness \
  --config benchmarks/config.example.yaml \
  --seed $CORTEX_SEED

# verify a published run reproduces bit-identically
python -m benchmarks.harness \
  --verify benchmarks/results/<run-id>.json
Reproducibility protocol — deterministic re-execution and closure verification.
9.4

Documentation Portal

Structured entry points into the research portal, organised by reading path — from the research programme and methodology, through the framework architecture and modules, to the experimental evaluation and external reference list.

Getting Started
Framework
Evaluation
About