agent-os-kernel 1.1.0__py3-none-any.whl → 1.2.0__py3-none-any.whl

modules/cmvk/.gitignore

@@ -0,0 +1,147 @@
+# =============================================================================
+# Cross-Model Verification Kernel - .gitignore
+# =============================================================================
+
+# -----------------------------------------------------------------------------
+# Python
+# -----------------------------------------------------------------------------
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+*.pyo
+*.pyd
+
+# Distribution / packaging
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# -----------------------------------------------------------------------------
+# Virtual environments
+# -----------------------------------------------------------------------------
+venv/
+env/
+ENV/
+.venv/
+.env.local
+
+# -----------------------------------------------------------------------------
+# IDE and Editors
+# -----------------------------------------------------------------------------
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+.DS_Store
+Thumbs.db
+*.sublime-project
+*.sublime-workspace
+
+# -----------------------------------------------------------------------------
+# Jupyter Notebook
+# -----------------------------------------------------------------------------
+.ipynb_checkpoints/
+*.ipynb_checkpoints/
+
+# -----------------------------------------------------------------------------
+# Testing and Coverage
+# -----------------------------------------------------------------------------
+.pytest_cache/
+.coverage
+.coverage.*
+htmlcov/
+.tox/
+.nox/
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+coverage_html/
+
+# -----------------------------------------------------------------------------
+# Type checking
+# -----------------------------------------------------------------------------
+.mypy_cache/
+.dmypy.json
+dmypy.json
+.pytype/
+
+# -----------------------------------------------------------------------------
+# Logs and Traces
+# -----------------------------------------------------------------------------
+logs/
+*.log
+traces/
+
+# But include sample traces for demonstration
+!logs/traces/demo_*.json
+
+# -----------------------------------------------------------------------------
+# Environment variables
+# -----------------------------------------------------------------------------
+.env
+.env.local
+.env.*.local
+
+# -----------------------------------------------------------------------------
+# Experiments and Results
+# -----------------------------------------------------------------------------
+experiments/results/
+*.json.bak
+
+# Note: experiments/datasets/*.json files ARE included in git
+# These are benchmark datasets (HumanEval) that are small enough
+# to commit for convenience. See experiments/datasets/README.md
+
+# -----------------------------------------------------------------------------
+# Docker
+# -----------------------------------------------------------------------------
+.dockerignore
+
+# -----------------------------------------------------------------------------
+# Temporary files
+# -----------------------------------------------------------------------------
+/tmp/
+*.tmp
+*.bak
+*.temp
+
+# -----------------------------------------------------------------------------
+# Model outputs
+# -----------------------------------------------------------------------------
+outputs/
+checkpoints/
+
+# -----------------------------------------------------------------------------
+# Build and Release
+# -----------------------------------------------------------------------------
+.pypirc
+
+# Hatch
+.hatch/
+
+# -----------------------------------------------------------------------------
+# Documentation builds
+# -----------------------------------------------------------------------------
+site/
+docs/_build/

modules/cmvk/.pre-commit-config.yaml

@@ -0,0 +1,58 @@
+# Pre-commit hooks for CMVK
+# Install: pip install pre-commit && pre-commit install
+# Run all: pre-commit run --all-files
+
+default_language_version:
+  python: python3
+
+repos:
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v5.0.0
+    hooks:
+      - id: trailing-whitespace
+      - id: end-of-file-fixer
+      - id: check-yaml
+        args: [--unsafe]
+      - id: check-json
+      - id: check-toml
+      - id: check-added-large-files
+        args: ['--maxkb=1000']
+      - id: check-merge-conflict
+      - id: detect-private-key
+      - id: check-case-conflict
+      - id: mixed-line-ending
+        args: ['--fix=lf']
+
+  - repo: https://github.com/psf/black
+    rev: 24.10.0
+    hooks:
+      - id: black
+        args: ['--line-length=100']
+
+  - repo: https://github.com/pycqa/isort
+    rev: 5.13.2
+    hooks:
+      - id: isort
+        args: ['--profile=black', '--line-length=100']
+
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.9.2
+    hooks:
+      - id: ruff
+        args: ['--fix', '--exit-non-zero-on-fix']
+
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: v1.14.1
+    hooks:
+      - id: mypy
+        additional_dependencies: [types-PyYAML, types-requests]
+        args: ['--ignore-missing-imports']
+        exclude: ^(tests/|experiments/|notebooks/)
+
+  - repo: local
+    hooks:
+      - id: pytest-check
+        name: pytest-check
+        entry: python -m pytest tests/unit/ -x -q --no-header
+        language: system
+        pass_filenames: false

modules/cmvk/CHANGELOG.md

@@ -0,0 +1,146 @@
+# Changelog
+
+All notable changes to the Cross-Model Verification Kernel (CMVK) will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [1.1.0] - 2026-01-21
+
+### Changed
+- **BREAKING**: Renamed package from `src` to `cross_model_verification_kernel`
+  - All imports must now use `from cross_model_verification_kernel import ...`
+  - CLI entry point updated to `cross_model_verification_kernel.cli:app`
+- Reorganized documentation into `docs/` directory
+  - `ARCHITECTURE.md` → `docs/architecture.md`
+  - `SAFETY.md` → `docs/safety.md`
+  - `NEW_FEATURES.md` → `docs/features.md`
+  - `INNOVATION_LAYER.md` → `docs/innovation_layer.md`
+  - `FEATURE_3_TRACEABILITY.md` → `docs/traceability.md`
+- Reorganized tests into `tests/unit/` and `tests/integration/`
+- Moved example scripts to `examples/` directory
+- Moved paper-related files to `paper/` directory
+- Updated `.pre-commit-config.yaml` to v5.0.0 hooks
+- Modernized `.gitignore` with comprehensive patterns
+
+### Added
+- `src/cross_model_verification_kernel/__main__.py` for `python -m` support
+- `CONTRIBUTING.md` with development guidelines
+- `requirements-dev.txt` for contributors
+- `tests/conftest.py` with shared pytest fixtures
+- Merged `docs/getting_started.md` (combined GETTING_STARTED + QUICKSTART)
+- Professional README.md with badges, benchmarks, and quick start guide
+
+### Removed
+- Redundant files: `IMPLEMENTATION_COMPLETE.md`, `IMPLEMENTATION_SUMMARY.md`,
+  `PROJECT_COMPLETE.md`, `NEXT_STEPS.md`, `LAUNCH_CHECKLIST.md`, `hf_readme.md`
+- Old `sys.path` hacks in example files (now uses proper package imports)
+
+## [Unreleased]
+
+### Added
+- **LaTeX paper** (`paper/cmvk_neurips.tex`)
+  - NeurIPS 2025 format with full methodology
+  - BibTeX references (`paper/references.bib`)
+  - Publication-ready tables and algorithm
+- **SVG figures** for paper/presentations
+  - Architecture diagram (`paper/figures/architecture.svg`)
+  - Results bar chart with error bars (`paper/figures/results_bar.svg`)
+  - Ablation study chart (`paper/figures/ablation.svg`)
+  - Figure generation script (`paper/generate_figures.py`)
+- **LICENSE** file (MIT)
+- Anthropic Claude verifier support (`AnthropicVerifier`)
+- CLI interface with `cmvk` command
+  - `cmvk run` - Run verification on a task
+  - `cmvk config` - Manage configuration
+  - `cmvk benchmark` - Run benchmark experiments
+  - `cmvk visualize` - Visualize trace files
+  - `cmvk models` - List available models
+- Reproducibility controls with seed configuration
+- `pyproject.toml` for modern Python packaging
+- GitHub Actions CI/CD pipeline
+- Docker multi-stage build with production/sandbox/development targets
+- Per-agent temperature controls
+- Benchmark results table in README (with actual numbers, means±std, p-values)
+- HuggingFace Hub integration (`src/tools/huggingface_upload.py`)
+  - Upload datasets, traces, and results
+  - Auto-generate dataset cards
+- Statistical analysis utilities (`src/tools/statistics.py`)
+  - Welch's t-test, Wilcoxon signed-rank test
+  - Confidence intervals, bootstrap CI
+  - Effect size (Cohen's d)
+  - Results table formatting
+- Reproducible experiment runner (`experiments/reproducible_runner.py`)
+  - Hardware/runtime stats collection
+  - Deterministic execution with seeds
+  - Complete experiment logging
+- Safety and ethics documentation (`SAFETY.md`)
+  - Sandbox security guidelines
+  - Prompt injection defenses
+  - Dual-use considerations
+  - Responsible disclosure policy
+- Comprehensive test suite for new components
+  - `test_anthropic_verifier.py` - Anthropic adapter tests
+  - `test_cli.py` - CLI tests
+  - `test_reproducibility.py` - Seed/reproducibility tests
+- Pre-commit configuration (`.pre-commit-config.yaml`)
+- Architecture diagrams (`docs/DIAGRAMS.md`)
+  - Mermaid diagrams for architecture, verification loop, state machine
+  - ASCII art diagrams for terminals
+  - Export instructions for draw.io and LaTeX
+  - Color scheme reference
+- arXiv submission checklist (`ARXIV_CHECKLIST.md`)
+  - NeurIPS/ICLR/ICML requirements
+  - LLM usage disclosure templates
+  - Reproducibility checklist
+  - Anonymization guidelines
+- Ablation experiment runner (`experiments/ablation_runner.py`)
+  - Predefined configurations for baselines, cross-model, loop depth
+  - Statistical aggregation across runs
+  - LaTeX table generation
+- Results visualization (`experiments/visualize_results.py`)
+  - ASCII bar charts and tables
+  - SVG chart generation (no dependencies)
+  - LaTeX table export
+- Enhanced PAPER.md
+  - Formal problem definition with mathematical notation
+  - Blind spot reduction theorem
+  - Computational complexity analysis
+  - Expanded related work with positioning table
+  - Complete reference list (14 citations)
+
+### Changed
+- Pinned all dependency versions in `requirements.txt`
+- Updated README with installation via pip, CLI usage, and Docker instructions
+- Improved configuration handling with seed support
+- Dockerfile now uses multi-stage build with separate targets
+
+### Fixed
+- N/A
+
+## [1.0.0] - 2024-01-21
+
+### Added
+- Initial release of Cross-Model Verification Kernel
+- OpenAI Generator agent (GPT-4o, GPT-4 Turbo, o1 models)
+- Gemini Verifier agent (Gemini 1.5 Pro/Flash)
+- Graph of Truth state machine for loop prevention
+- Prosecutor Mode for hostile test generation
+- Trace logging and visualization system
+- HumanEval dataset integration
+- Lateral thinking via strategy banning
+- Sandbox code execution
+- Configuration via YAML files
+- Basic test suite
+- Research paper draft (PAPER.md)
+
+### Security
+- Basic sandbox isolation for code execution
+- API key handling via environment variables
+
+---
+
+## Version History
+
+- **1.0.0**: Initial research release with OpenAI + Gemini support
+- **1.1.0** (upcoming): Anthropic support, CLI, improved reproducibility

modules/cmvk/CITATION.cff

@@ -0,0 +1,48 @@
+# CITATION.cff - Citation File Format for academic citations
+# https://citation-file-format.github.io/
+# This file enables "Cite this repository" on GitHub
+
+cff-version: 1.2.0
+title: "Cross-Model Verification Kernel (CMVK)"
+message: "If you use this software in your research, please cite it as below."
+type: software
+authors:
+  - family-names: "Siddique"
+    given-names: "Imran"
+    orcid: "https://orcid.org/0000-0000-0000-0000"  # Update with your ORCID
+repository-code: "https://github.com/imran-siddique/cross-model-verification-kernel"
+url: "https://github.com/imran-siddique/cross-model-verification-kernel"
+license: MIT
+version: "0.1.0"
+date-released: "2026-01-01"
+keywords:
+  - verification
+  - ai-safety
+  - hallucination-detection
+  - drift-detection
+  - adversarial
+  - multi-model
+  - code-generation
+  - llm
+abstract: >-
+  CMVK is an adversarial multi-model verification framework that addresses
+  the "grading your own homework" fallacy in AI systems. By employing
+  strategic model diversity with separate Generator and Verifier roles,
+  CMVK reduces correlated blind spots and detects hallucinations that
+  self-verification would miss.
+
+# Preferred citation for the paper (update when published)
+preferred-citation:
+  type: conference-paper
+  authors:
+    - family-names: "Siddique"
+      given-names: "Imran"
+  title: "Cross-Model Verification Kernel: Adversarial Multi-Model Code Generation with Blind Spot Reduction"
+  year: 2026
+  collection-title: "Advances in Neural Information Processing Systems"
+  collection-type: proceedings
+  conference:
+    name: "NeurIPS 2026"
+  # Update these fields when the paper is accepted/published:
+  # url: "https://proceedings.neurips.cc/paper/..."
+  # doi: "10.xxxx/..."

modules/cmvk/CONTRIBUTING.md

@@ -0,0 +1,229 @@
+# Contributing to CMVK
+
+Thank you for your interest in contributing to the Cross-Model Verification Kernel! This document provides guidelines and instructions for contributing.
+
+## Development Setup
+
+### 1. Fork and Clone
+
+```bash
+git clone https://github.com/YOUR_USERNAME/cross-model-verification-kernel.git
+cd cross-model-verification-kernel
+```
+
+### 2. Create a Virtual Environment
+
+```bash
+python -m venv .venv
+source .venv/bin/activate  # Linux/Mac
+# or
+.venv\Scripts\activate     # Windows
+```
+
+### 3. Install Development Dependencies
+
+```bash
+pip install -e ".[dev]"
+# or
+pip install -e ".[all]"  # includes notebooks, sandbox, etc.
+```
+
+### 4. Install Pre-commit Hooks
+
+```bash
+pre-commit install
+```
+
+This will automatically run linting and formatting checks before each commit.
+
+## Code Standards
+
+### Style
+
+- **Formatter**: [Black](https://black.readthedocs.io/) with 100 character line length
+- **Linter**: [Ruff](https://docs.astral.sh/ruff/) for fast, comprehensive linting
+- **Import Sorting**: [isort](https://pycqa.github.io/isort/) with Black-compatible profile
+- **Type Hints**: Required for all public functions
+
+### Running Checks Manually
+
+```bash
+# Format code
+black src/ tests/
+
+# Sort imports
+isort src/ tests/
+
+# Run linter
+ruff check src/ tests/ --fix
+
+# Type checking
+mypy src/
+
+# Run all checks
+pre-commit run --all-files
+```
+
+## Testing
+
+### Running Tests
+
+```bash
+# Run all tests
+pytest tests/ -v
+
+# Run with coverage
+pytest tests/ --cov=src/cross_model_verification_kernel --cov-report=html
+
+# Run specific test file
+pytest tests/test_kernel.py -v
+
+# Run tests matching a pattern
+pytest tests/ -k "test_verification" -v
+
+# Skip slow/integration tests
+pytest tests/ -m "not slow and not integration"
+```
+
+### Writing Tests
+
+- Place tests in the `tests/` directory
+- Use descriptive test names: `test_kernel_rejects_same_model_generator_verifier`
+- Use pytest fixtures for common setup
+- Mock external API calls (don't make real API calls in tests)
+
+Example:
+
+```python
+import pytest
+from unittest.mock import MagicMock
+
+from cross_model_verification_kernel import VerificationKernel
+
+def test_kernel_initialization():
+    """Test that kernel initializes with valid configuration."""
+    generator = MagicMock()
+    verifier = MagicMock()
+
+    kernel = VerificationKernel(generator=generator, verifier=verifier)
+
+    assert kernel.generator == generator
+    assert kernel.verifier == verifier
+```
+
+## Commit Messages
+
+We follow [Conventional Commits](https://www.conventionalcommits.org/):
+
+```
+<type>(<scope>): <description>
+
+[optional body]
+
+[optional footer(s)]
+```
+
+### Types
+
+- `feat`: New feature
+- `fix`: Bug fix
+- `docs`: Documentation changes
+- `style`: Code style changes (formatting, etc.)
+- `refactor`: Code refactoring
+- `test`: Adding or updating tests
+- `chore`: Maintenance tasks
+
+### Examples
+
+```
+feat(kernel): add strategy banning mechanism
+fix(verifier): handle empty response from Gemini API
+docs(readme): update installation instructions
+test(agents): add unit tests for AnthropicVerifier
+```
+
+## Pull Request Process
+
+1. **Create a Branch**
+   ```bash
+   git checkout -b feat/your-feature-name
+   ```
+
+2. **Make Changes**
+   - Write code
+   - Add tests
+   - Update documentation if needed
+
+3. **Run Checks**
+   ```bash
+   pre-commit run --all-files
+   pytest tests/ -v
+   ```
+
+4. **Commit and Push**
+   ```bash
+   git add .
+   git commit -m "feat(scope): description"
+   git push origin feat/your-feature-name
+   ```
+
+5. **Open a Pull Request**
+   - Provide a clear description of changes
+   - Link any related issues
+   - Ensure CI passes
+
+## Areas for Contribution
+
+### High Priority
+
+- **New Model Providers**: Add support for LLaMA, Mistral, Cohere, etc.
+- **Enhanced Prompts**: Improve adversarial verification prompts
+- **Bug Fixes**: Check open issues for bugs
+
+### Research Extensions
+
+- **New Datasets**: Add support for other benchmarks (MBPP, LogicGrids, etc.)
+- **Ablation Studies**: Run experiments with different configurations
+- **Graph Memory Improvements**: Better state management algorithms
+
+### Documentation
+
+- **Tutorials**: Step-by-step guides for common use cases
+- **API Reference**: Docstrings and type hints
+- **Examples**: More example scripts
+
+## Project Structure
+
+```
+src/cross_model_verification_kernel/
+├── __init__.py          # Package exports
+├── __main__.py          # python -m support
+├── cli.py               # CLI commands
+├── core/
+│   ├── kernel.py        # Main verification loop
+│   ├── graph_memory.py  # Graph of Truth
+│   └── types.py         # Data classes
+├── agents/
+│   ├── base.py          # Abstract base class
+│   ├── generator_openai.py
+│   ├── verifier_gemini.py
+│   └── verifier_anthropic.py
+├── tools/
+│   ├── sandbox.py       # Code execution
+│   └── visualizer.py    # Trace replay
+└── datasets/
+    └── humaneval_loader.py
+```
+
+## Questions?
+
+- **Issues**: Open a GitHub issue for bugs or feature requests
+- **Discussions**: Use GitHub Discussions for questions
+
+## License
+
+By contributing, you agree that your contributions will be licensed under the MIT License.
+
+---
+
+Thank you for contributing! 🚀