02 E-COMMERCE Regulated sectors High

E-commerce — Monolith to Microservices Migration

Marketplace with a 200K-line Django monolith progressively extracts services with 60% AI-generated code.

◻ intake ★★★

△ architect ★★★

◇ vigil ★★☆

⬡ licit ★★☆

Architecture diagram

Context

A marketplace with a 200K-line Django monolith progressively extracts services. 60% of new code is generated by AI agents. The team needs to track which microservices were AI-generated for compliance and accountability.

Flow with 4 tools

◻ Phase 01 — intake

Spec per bounded context

Generates separate specs per bounded context from Jira and existing source code.

◻ intake

intake init "Extract Users Service" \
  --source jira://SHOP/label:users-extraction \
  --source src/monolith/apps/users/ \
  --format architect

△ Phase 02 — architect

Parallel implementation

Extracts multiple services simultaneously with agents working in parallel.

△ architect

architect parallel \
  --task "Extract Users Service per spec/" \
  --task "Extract Catalog Service per spec/" \
  --task "Implement API Gateway with Traefik"

◇ Phase 03 — vigil

Per-service verification

Scans each service independently and generates consolidated SARIF reports.

◇ vigil

for svc in services/*/; do
  vigil scan "$svc" --format sarif --output "reports/vigil-$(basename $svc).sarif"
done

⬡ Phase 04 — licit

Migration provenance tracking

Tracks which microservices were AI-generated and generates agent configuration changelog.

⬡ licit

licit trace --since "2024-06-01"
licit changelog
licit report --format html --output reports/migration-compliance.html
licit verify --min-score 70

Why licit is critical here

During a massive migration where AI agents generate 60% of new code, licit tracks exactly which services have AI-generated code, which models were used, and whether guardrails were modified during the process.