clm-kernel

Card Lifecycle Manager (CLM) Kernel — Sovereign Functional Agentic Operating System.

clm-kernel is the foundational Python implementation of the Cubical Logic Model (CLM), providing a stratified six-layer architecture (layer0 … layer5) for sovereign, functional, deterministically auditable agentic execution.

Layer model (DOC-02). This package implements the six logical strata defined in docs/architecture/six_layers_architecture.md and docs/spec/20-architecture/system-architecture.md. The authoritative stratum ↔ package mapping is docs/sprints/13_documentation-reconciliation/DOC-02-layer-model-reconciliation.md.

Architectural Layers

• Layer 0: Stateless Kenotic Core (Microkernel) (clm/layer0/)
  • Pure algebraic primitives, AST parsers, codecs, hashing, and static schemas
  • Content-addressable identity and MCard primitives
• Layer 1: Colored Petri Net (CPN) Virtual Machine (clm/layer1/)
  • Coloured Petri Net (CPN) virtual machine, token passing, transition firings
  • Incidence-matrix / Noetherian invariant solving; Baldwin operator algebra
• Layer 2: Content-Addressable Storage (CAS) & Cordis Fiber Substrate (clm/layer2/)
  • Tri-Database storage, CAS/Merkle substrate, Cordis spatiotemporal fibers
  • Effect isolation, savepoints (SavepointGuard), and cleanup scopes
• Layer 3: Distributed Network Gateway & Multi-Agent Coordination (clm/layer3/)
  • Reticulum mesh transport/discovery, Satori speech-act protocol, handle routing
• Layer 4: Human Interaction Surface & Conversational Programming (clm/layer4/)
  • Conversational gateways, presentation membranes, EOA sponsor signing, CLI (clm, clmEval)
• Layer 5: System Orchestration, Evaluator Engine & Playbooks (clm/layer5/)
  • Continuous-learning stratum ($\Gamma$), epistemic fitness ($\Delta E$), telemetry hooks

Installation

pip install clm-kernel

CLI Usage

# Display CLI help
clm --help

# Evaluate an action or workflow manifest
clm evaluate --file workflow.yaml

# Initialize TriDatabase storage
clm init-tridb --path ./data

Python API Quick Start

from clm.layer0.hash import prefixed_hash
from clm.layer0.mcard import MCardPayload, create_mcard, verify_mcard_integrity

# Construct a content-addressed MCard (BLAKE3) and verify its integrity
card = create_mcard(
    uri="mcard://example/greeting",
    payload=MCardPayload(kind="Text", value="payload data"),
    author="did:clm:example",
)

print(f"Content hash: {prefixed_hash(b'payload data')}")  # blake3:<hex>
print(f"Card hash:    {card.hash}")
assert verify_mcard_integrity(card)

License

MIT License. See LICENSE for details.

Monorepo & Documentation

Source code, formal proofs, and multi-language kernels (Rust, JavaScript/TypeScript, WASM) are maintained in the main CLM Repository.