Skip to Content

Cookbook: Scaffolding, Developing, and Deploying Governed Enterprise Applications with DevEx Pack, CLI, and Vibe

This guide walks through an end-to-end journey in an enterprise environment: from AEP blueprint discovery → CLI scaffolding → Vibe agentic pair programming → 5-layer test pyramid validation → AEP production release & GitOps deployment.

1. Architectural Mental Model: Three-Layer Triad

In modern enterprise application engineering, systems are not written from raw unconstrained templates. Instead, applications are composed of three structured layers:

┌────────────────────────────────────────────────────────────────────────┐
│ [Layer 1] 5 Standard DevEx Packs (Technology stack, SDKs & Sandbox)     │
│   · aix-governed-admin-pack (Governed CRUD) · aix-agentic-workbench-pack│
│   · aix-lakehouse-dataproduct-pack        · aix-feishu-copilot-pack    │
│   · aix-autonomous-pipeline-pack                                       │
└───────────────────────────────────┬────────────────────────────────────┘
                                    │ Binds to
┌───────────────────────────────────▼────────────────────────────────────┐
│ [Layer 2] 5 Generic UI Shell Archetypes (Interaction flows & layout)    │
│   · admin-datatable (Admin Tables) · split-workbench (Dual-pane)       │
│   · conversational-chat (Copilot)  · metrics-dashboard (KPI Dashboard) │
│   · feishu-card (Interactive Card)                                     │
└───────────────────────────────────┬────────────────────────────────────┘
                                    │ Business Context Injection
┌───────────────────────────────────▼────────────────────────────────────┐
│ [Layer 3] Enterprise Business Instances                                │
│   • Cross-Border Channel Commission Rules Admin (with Diff Auditing)   │
│   • Amazon PPC Ad ACOS Optimizer Copilot (with Circuit Breaker)        │
│   • Ocean Demurrage Triage Workbench (with Last Free Day Alerts)       │
└────────────────────────────────────────────────────────────────────────┘

2. Step 1: Discovering Foundation Packs & Blueprints

Check all available enterprise packs and certified blueprints via the CLI:

# 1. View 5 standard generic DevEx Packs
aix aep packs

Output:

📦 Registered DevEx Packs (5 Enterprise Standards):
  • aix-governed-admin-pack (v1.0.0)
    Name: Governed Enterprise Admin & CRUD DevEx Pack | Connector: aix_admin
    Capabilities: admin-crud-datatable, config-dictionary-desk, audit-trail-recorder
    Description: Enterprise-grade governed CRUD and admin suites with built-in audit diff snapshots.

  • aix-agentic-workbench-pack (v1.0.0)
    Name: Governed Agentic Workbench DevEx Pack | Connector: aix_agent
    Capabilities: agentic-split-workbench, conversational-copilot, mcp-tool-integrator
    Description: Human-in-the-loop AI workbench and conversational copilots.
  ...
# 2. View available scenario templates
aix aep templates

3. Step 2: Scaffolding Projects via CLI

We demonstrate two core enterprise application archetypes:

Scenario A: Governed Admin Application (CRUD + Diff Audit Trail)

Business Context: Operations teams maintain commission ratios and minimum gross margin guardrails across Amazon, TikTok Shop, and Walmart. Every change requires immutable before/after diff snapshots.

aix aep init my-channel-admin \
  --template tpl-sn-channel-rules-admin \
  --pack aix-governed-admin-pack \
  --project proj-commercial-tower \
  --dir ./my-channel-admin

Output:

✨ [AEP] Successfully initialized governed application project!
  • App ID:     app-sn-channel-rules-admin
  • Directory:  ./my-channel-admin
  • DevEx Pack: aix-governed-admin-pack
  • Blueprint:  tpl-sn-channel-rules-admin
  • Archetype:  Governed Admin CRUD (Immutable Diff Audit)

Scenario B: Governed Agentic Copilot (AI + Sandbox + Evidence)

Business Context: Ad managers require an AI assistant monitoring Amazon PPC campaign ACOS. When ACOS degrades beyond 25%, it safely throttles bids and records signed structured evidence.

aix aep init my-ad-optimizer \
  --template tpl-sn-amazon-ad-optimizer \
  --pack aix-agentic-workbench-pack \
  --project proj-marketing-growth \
  --dir ./my-ad-optimizer

4. Step 3: Exploring Generated Project Structure

Inspect the generated project directory:

cd my-channel-admin
tree .
.
├── aep-app.yaml          # AEP governed application contract declaration
├── pyproject.toml        # Dependencies and test configuration
├── src/
│   ├── admin.py          # GovernedRecordStore: CRUD logic with Before/After Diff snapshots
│   └── app.py            # GovernedAdminApp: Tenant isolation, RBAC checks & mutation gates
└── tests/
    ├── test_admin.py     # Business and safety unit tests
    └── test_cdd_contracts.py # CDD contract compliance tests

1. aep-app.yaml Contract Manifest

schema_version: aep.governed-application/v1
id: app-sn-channel-rules-admin
name: Cross-Border Channel Commission & Pricing Rules Admin Desk
organization_id: org-production
project_id: proj-commercial-tower
template_id: tpl-sn-channel-rules-admin
pack_id: aix-governed-admin-pack
lifecycle: draft
technology_stack: aix-admin
governance:
  audit_trail: immutable_diff
  mutation_approval_gate: true
  two_person_rule: true

2. Diff Audit Implementation in src/admin.py

def update_record(self, record_id: str, patch: dict, actor: str) -> dict:
    old_data = dict(self._records[record_id]["data"])
    new_data = dict(old_data)
    new_data.update(patch)
 
    audit_entry = {
        "action": "UPDATE",
        "record_id": record_id,
        "actor": actor,
        "diff": {"before": old_data, "after": new_data},  # Explicit field-level diff
        "timestamp": time.time(),
        "verified": True,
    }
    self._audit_log.append(audit_entry)
    return self._records[record_id]

5. Step 4: Local Test Pyramid Verification (Pytest)

Run tests locally to verify adherence to all AEP platform security and business contracts:

pytest -v

Output:

tests/test_admin.py::test_create_record_generates_diff_audit PASSED      [ 25%]
tests/test_admin.py::test_update_record_tracks_diff PASSED               [ 50%]
tests/test_admin.py::test_invalid_record_rejected PASSED                 [ 75%]
tests/test_admin.py::test_governed_admin_app_tenant_isolation PASSED    [100%]
tests/test_cdd_contracts.py::test_aep_application_manifest_conformance PASSED

============================== 5 passed in 0.08s ===============================

All business validation, audit snapshot generation, and aep-app.yaml compliance tests pass 100%.

6. Step 5: AI-SDLC Pair Programming with Vibe

Developers can summon specialized engineering agents via vibe to extend features under contract guardrails:

# Summon an engineering agent to connect to production schemas and expand test suites
aix vibe run "Integrate real channel commission schema and add alert tests for >30% fee rates" --unit my-channel-admin

Vibe follows CDD contract specifications, practicing Test-Driven Development (TDD) to implement logic and automatically run regression test suites.

7. Step 6: Submitting to AEP Governance & Production Deployment

Once local verification completes, submit metadata and test evidence to the AEP control plane:

# Submit AEP application release application
aix aep release apply --app-id app-sn-channel-rules-admin
 
# Inspect release status and cryptographic evidence signatures
aix aep release status --app-id app-sn-channel-rules-admin

Following approval, the AEP GitOps controller deploys workloads directly to the enterprise Kubernetes cluster and makes them accessible in the enterprise portal.