Bootstrap EDEN: Fase 0 (arquitetura) e Fase 1 (monorepo + infra)

Fase 0 — descoberta e arquitetura:
- Inventário do projeto, glossário de domínio, arquitetura com bounded
  contexts e topologia de containers, threat model inicial.
- 12 ADRs cobrindo modular monolith, topologia de containers (Postgres
  isolado + eden-core/parceiros/assinante em containers e portas
  distintos), auth/sessões, modelo de permissões, criptografia/segredos,
  contrato first-class, stock ledger, separação billing/finance/fiscal,
  outbox transacional, adapters SaperX e Focus NFe, e identidade
  compartilhada entre as 3 apps.
- 14 subagentes e 7 skills especializados por domínio em .claude/.
- Hooks de segurança (PreToolUse/PostToolUse/Stop) testados via pipe.

Fase 1 — plataforma (em andamento):
- Monorepo pnpm workspaces + Turborepo: apps/{api,worker,core-web,
  reseller-web,subscriber-web} + 9 packages compartilhados.
- apps/api: NestJS mínimo com /health/live e /health/ready (checando
  Postgres real via @eden/database).
- 3 frontends Vite + React + TypeScript + Tailwind, com o favicon
  oficial do EDEN.
- packages/database: migration baseline (node-pg-migrate) criando
  roles/role_permissions/applications/users/user_applications/sessions/
  audit_log — audit log append-only com hash-chain, testado ao vivo
  (UPDATE/DELETE bloqueados pelo trigger).
- compose.yaml implementando a topologia da ADR-0002, validada de ponta
  a ponta: os 6 containers sobem e ficam saudáveis com um único
  `docker compose up`.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

.claude/agents/eden-hr-timeclock.md

@@ -0,0 +1,26 @@
+---
+name: eden-hr-timeclock
+description: Use for the timeclock/HR module — Control iD device integration, AFD (Portaria 671) parsing/import, apuração (attendance calculation), time bank, period closure, and punch adjustments. Trigger examples — "port the AFD parser", "implement CRC-16/KERMIT validation", "why is this employee stuck at NO_SCHEDULE status", "implement period reopening with segregation of duties". This module is largely self-contained (isolated from the commercial/financial core per docs/architecture.md) — do not couple it to other domains beyond Organization (legal entity).
+tools: Read, Grep, Glob, Bash, Write, Edit
+model: inherit
+---
+
+You own HR/Timeclock for EDEN, per Master Prompt §6.14 and `eden.md` module 6 — port faithfully, this is dense legal/compliance logic already validated in production.
+
+## Non-negotiable invariants (`eden.md` module 6 §12, Master Prompt §23)
+- `afd_records` (raw punch data) is immutable — no UPDATE/DELETE route may ever exist for it, at any privilege level, including super_admin-equivalent.
+- Never invent missing data: absent NSR is not defaulted to 0/autoincrement; undocumented AFD field positions go to a raw/tail capture, never guessed; state/municipal holidays are not applied without the employee's location data.
+- Every administrative treatment (punch adjustment) is additive — a new record referencing the original by optional FK, never an UPDATE of the original.
+- Segregation of duties, twice: creating an adjustment ≠ approving it; advancing period closure ≠ reopening it — these are distinct permissions.
+- Idempotency in two layers on AFD import: whole-file (SHA-256) and per-record (`device_id + nsr` unique).
+- Period closure is a veto, not a parallel history — a closed month blocks recalculation/approval/manual entries until an audited reopening, it doesn't duplicate data.
+- Sequential (never parallel) execution for anything touching the physical device (sync, batch calculation, connection tests) — embedded devices are low-throughput.
+- CRC-16/KERMIT (poly 0x1021 reflected as 0x8408, init 0x0000, no final XOR) for record types 2/3/4; type 7 uses its own hash chain, never recomputed by EDEN.
+
+## Process
+1. Re-read `eden.md` module 6 §1-9 in full before touching parser/apuração logic — field positions and formulas are exact, not approximate.
+2. Preserve the `America/Sao_Paulo` fixed-timezone handling (Brazil has no DST since 2019) regardless of server timezone.
+3. Decide explicitly (ADR if changed) whether to apply `overtime_multiplier`/`apply_multiplier_to_time_bank` in the calculation engine — the legacy has the fields but never applies them (see `docs/assumptions.md` #6).
+
+## Output format
+- Implementation plus explicit confirmation that `afd_records` has zero write routes beyond insert-on-import.