Table of Contents

Tutorials

AsiBackbone Learning tutorials are problem-first. They begin with an architectural problem, expose a failure mode or limitation, introduce a pattern, and connect the teaching example to runnable evidence and fuller implementations.

The goal is understanding—not framework adoption.

Code scope: Tutorial snippets and companion projects are Learning-owned teaching models unless a section is explicitly labeled AsiBackbone 7.0 API. For exact current namespaces and syntax, use the AsiBackbone 7.0 Compatibility and API Boundary.

Learning Path at a Glance

Step Tutorial Difficulty Boundary added
1 Decision Before Execution Beginner Evaluation is separated from protected execution
2 Policy Context and Explicit Decision Outcomes Beginner Decision facts and outcomes become explicit
3 Decision Receipts and Acknowledgment Intermediate Decision receipts, acknowledgment, and later lifecycle evidence remain distinct from authority
4 Scoped Capability and Host-Owned Execution Intermediate Execution authority becomes narrow, temporary, and host-validated
5 Governed AI Tool Gateway Intermediate AI proposal is composed with host-owned context, policy, authority, and execution

All five are currently classified as Canonical Pattern material. Difficulty describes conceptual complexity, not production readiness.

If you already know authorization, ABAC, capability security, workflow, audit/provenance, or reference-monitor concepts, use Terminology and Established Architecture Concepts to map that vocabulary to the terms used here.

How a Tutorial Works

A typical tutorial follows this progression:

Problem
   ↓
Common or naive implementation
   ↓
Failure mode or limitation
   ↓
Architectural pattern
   ↓
Minimal teaching example
   ↓
Tradeoffs and alternatives
   ↓
Working repository example

Each foundational tutorial also includes:

  • a Pattern Card for fast orientation,
  • an observable invariant that carries into samples or tests,
  • tradeoffs and simpler alternatives,
  • and a Check Your Understanding checklist focused on what you should be able to explain or demonstrate.

The checklist is not a score or certification.

The Five Foundations

1. Decision Before Execution

Represent a consequential operation as proposed intent, evaluate it, and produce an explicit decision before the host performs the side effect.

Core idea: intent, authorization, governance decision, execution, and evidence should not collapse into one opaque operation.

A proposed action should become a governed decision before it becomes real-world execution.

2. Policy Context and Explicit Decision Outcomes

Represent the facts used by policy explicitly and return outcomes that describe what happens next rather than reducing every decision to a boolean.

Core ideas: actor/resource/operation/environment context, context snapshots, stable reason codes, policy identity, determinism, and decision composition.

3. Decision Receipts and Acknowledgment

Pause a consequential operation for explicit acknowledgment, resume through a governed boundary, and preserve structured evidence of the decision path.

Core ideas: response binding, expiration, replay, re-evaluation, acknowledgment versus override, correlation, and durable evidence boundaries.

4. Scoped Capability and Host-Owned Execution

Keep approval from becoming broad standing authority by issuing and validating short-lived, narrowly scoped execution authority at the host boundary.

Core ideas: subject/operation/resource/audience binding, time bounds, replay, revocation, current-state validation, and host-owned execution.

5. Governed AI Tool Gateway

Compose the first four patterns around AI-proposed tool execution while keeping authoritative context, credentials, policy, and real-world effects under host control.

AI proposal
   ↓
Host-owned context
   ↓
Governance decision
   ↓
Acknowledgment when required
   ↓
Scoped capability
   ↓
Execution-boundary validation
   ↓
Host-owned tool execution
   ↓
Decision receipt

The model may propose. The host retains execution authority.

Continue into Practice

Tutorials are the explanation layer. The broader learning path is:

Tutorial
   ↓
Executable Sample
   ↓
Hands-On Lab
   ↓
Working Repository

After a tutorial:

The five tutorials form the initial governed-execution curriculum, but they are meant to be questioned, simplified, adapted, or rejected when another design better fits the problem.

Read it. Run it. Question it. Improve it.

Edit this page