Module 15 — Capstone: Mini Open Engineering Kernel

Goal

Ship a working domain kernel that Python applications can import:

from open_engineering_kernel import (
    Identifier,
    validate_manifest,
    evaluate_rule,
)

Architecture

Python application
       │
       ├── AI / agents / workflows
       ▼
   Python API
       │
      PyO3
       │
       ▼
┌───────────────────────────┐
│    Open Engineering       │
│       Mini Kernel         │
│  Identifier · Manifest    │
│  Rule · Validation        │
└───────────────────────────┘
            │
           Rust

Requirement 1 — Identifier

id = Identifier.parse("oe.pico.lamp")
print(id.namespace)  # open-engineering
print(id.kind)       # pico
print(id.name)       # lamp

Invalid input must raise InvalidIdentifierError.

Requirement 2 — Manifest

result = validate_manifest({
    "kind": "pico",
    "name": "lamp",
})
assert result["valid"] is True
assert result["manifest"]["identifier"] == "oe.pico.lamp"

Requirement 3 — Rule

rule = {
    "id": "nod-when-ready",
    "when": {"op": "eq", "key": "intent", "value": "nod"},
    "then": {"op": "emit", "event": "nod"},
}
result = evaluate_rule(rule, {"intent": "nod"})
assert result["matched"] is True

Packaging

maturin build --release
# then install the resulting wheel into a clean environment

Definition of done (learner)

Rust tests pass (cargo test)
Python API imports cleanly
Identifier, Manifest, and Rule examples succeed
Wheel builds and installs in a fresh environment
You can explain why each piece lives on the Rust side of the boundary

Principle

One semantic core, multiple interfaces.