Module Reference

ApexLedger is organized into eight modules (M1–M8), each implemented end-to-end — schema → RLS → PL/pgSQL RPCs → API → tests — and all of them posting into one general ledger. This page maps the territory; the daily flows have their own guides.

The module map

Payroll (M2)

Indonesian payroll as data, not spreadsheets:

• Calculate — per employee, per period. The RPC joins the company calendar (working days), applies unpaid-day deductions, overtime (hourly rate = base ÷ 173, rounded first then multiplied), BPJS components from seeded rate config, and the official TER bracket table (PMK 168/2023 — 125 brackets) against the employee's PTKP category.
• Approve — posts the accrual journal: Dr Earnings expense / Cr Deductions payable + Cr AP Gaji.
• Disburse — posts Dr AP Gaji / Cr cash-bank. Per-employee math is auditable from the Entries drill-down.
• Guards: ENTRY_ALREADY_EXISTS (double calculate), INCOMPLETE_CALCULATION (approve early), CALENDAR_NOT_PROVISIONED, PERIOD_NOT_APPROVED.

Inventory & Work Orders (M3)

• Costing — moving average (recomputed on every receipt: (old_avg × old_qty + line_total) / new_qty) or FIFO (burns oldest cost layers); FEFO for expiry-dated items (nearest-expiry first, expired lots skipped).
• No negative stock — aggregated availability is checked before any lot burn; drift between lot table and aggregate raises LOT_STOCK_DRIFT and blocks the issue.
• Transfers — atomic issue + receive between warehouses of the same entity; cross-entity transfers are rejected.
• Work orders — BOM (components + waste %) + labor + FOH rate (overhead ÷ capacity). Completing consumes components pro-rata to yield, computes COGM = material + labor + FOH, breaks material down per inventory account, and posts Dr Finished Goods / Cr components / Cr labor / Cr FOH.

Treasury (M6)

• Kasbon (employee advances) — request → submit → the same dynamic approval engine as procurement stamps the required role → approve → disburse → settle. Settlement splits automatically: used ≤ amount posts expenses; leftover cash returns to the bank; overage becomes an additional employee claim. The receivable clears to zero.
• Bank reconciliation — import statement lines, then Auto-match: exact amount within ±3 days matches AR/AP payments. Unmatched lines stay flagged for manual review.
• Cash-flow forecast — read-only weekly projection: AR due dates (inflow) vs AP due dates + pending kasbon (outflow).

Fixed Assets (M7)

• Register — cost, salvage, useful life (months), straight-line or declining balance. Codes FA-YYYY-XXXXXX; the acquisition journal auto-posts.
• Depreciation — monthly batch (built-in scheduler: 01:00 on day 1, for the previous month), one aggregated journal per run. Straight-line: (cost − salvage) ÷ life, capped so book value never drops below salvage. Fully depreciated assets flip to FULLY_DEPRECIATED.
• Dispose — sale / write-off / donation; the RPC posts the full sign-aware exit journal (accumulated depreciation + proceeds ± loss/gain vs cost). Codes stay for audit.

Budgeting & Analytics (M8)

• Cycle — create (account × department × month × amount) → approve (DEPT_HEAD_FA) → lock (SUPER_ADMIN only; frozen for normal revision).
• Revise — requires a reason; snapshots the before-state (JSONB), deletes and reinserts lines atomically, increments the revision number. Even locked budgets can be revised by SUPER_ADMIN — with the trail intact.
• Vs-actual — joins POSTED journal lines per fiscal year against budget lines: budgeted, actual, variance amount and %.
• Trend — REVENUE / EXPENSE movement over the last N months, normal-balance aware.

How the modules connect

Every module ends its flow by posting a journal into M1 — payroll accruals, depreciation batches, COGM entries, AP/AR invoices. The month-end close is simply the moment all those postings are verified and the period locks. One ledger, one truth.

What's next

• Installation — get a server running.
• docs/USER_FLOWS.md in the repository — every flow above, in full detail.