MFG - production system of record
One controlled
production record.
This page describes the intended design. No MFG code exists yet. It sets out how one controlled production record would turn scattered demand, materials, capacity, and shop-floor evidence into work orders, WIP truth, quality-gated completion, and auditable decisions.
production order
Assemble finished SKU
release readiness
- Production versionmBOM + routing snapshot
- Material readiness1 component short (WMS)
- Capacity readinesswork-center + shift
- Asset availabilityno EAM downtime
The designed capabilities
What the record is designed to hold.
These are the capabilities the blueprint designs into one production record - a short line each. None of it is built; every line describes intent, not running software.
Orders
Work orders
Designed to carry a production order through a controlled lifecycle, not a status flag.
Recipe
BOM + routing snapshots
Designed to freeze an mBOM and routing into a production version, so what was made stays known.
Readiness
Material + capacity readiness
Designed to check WMS stock, SCM supply, and work-center capacity before work reaches the floor.
Execution
WIP truth
Designed to capture operation confirmations, moves, yield, scrap, rework, and downtime as evidence.
Quality
Quality-gated completion
Designed so a failed mandatory checkpoint blocks completion until an authorised approver releases it.
Handoff
Output declaration
Designed to declare finished output with lot or serial traceability, then hand off to WMS and FMS.
The production lifecycle
One order, planned to completed.
MFG is designed to carry a make-to-stock order through one controlled sequence - each step writing append-only evidence to the same record, never a silent status flag. This is the intended design, not running software.
Designed - concept, not live
01
Planned
Version + readiness checked
02
Released
Planner releases the order
03
Staged
Materials staged from WMS
04
In progress
Operators confirm on the floor
05
Completed
Output declared, quality-gated
The problem it targets
The facts exist. The record does not.
- Demand, materials, and maintenance each live in their own tool
- Shop-floor events and finance records rarely agree
- No single record traces a finished lot back to what made it
SignalsDesigned control
Release only against approved readiness.
- Material readiness would read WMS on-hand and SCM incoming supply
- Capacity would read work-center calendars and EAM availability
- A failed mandatory quality checkpoint would block completion until approved
Shop floorWhere it would sit
Where MFG is designed to sit.
MFG is a blueprint, so this is where it is intended to sit in the landscape - not a measured comparison. ERP manufacturing suites and standalone MES keep their real strengths; MFG's angle is a production system of record built for one family fabric and agent-ready over MCP by design.
MFG has no running software, so nothing here is benchmarked. ERP manufacturing suites and standalone MES are mature and proven; MFG is blueprint v0.1 with no code yet. This table is designed positioning - how MFG is intended to sit - not a claim that it beats anything today.
The designed architecture
How it is designed to hold together.
The story above is the plain version. For evaluators, here is the intended design - a controlled make-to-stock lifecycle, an agent-ready seam for Arin over MCP, and a strict family boundary. None of it is built. Every panel below describes design intent, not running software.
The designed make-to-stock flow
The blueprint's first proof would be a make-to-stock production order driven by demand and constrained by materials, capacity, and quality. It is designed as a controlled lifecycle, not a status flag - each step would write append-only evidence, and corrections would be reversals, adjustments, or rework loops, never silent edits.
- 01version
Pick the recipe
A demand signal would select a production version - an mBOM snapshot plus a routing snapshot with operations, work centers, and quality checkpoints.
ProductionVersion
- 02readiness
Check before release
Material readiness would read WMS on-hand and SCM incoming supply; capacity would read work-center calendars and EAM availability. Shortages surface first.
MaterialReservationRequest
- 03release
Release with a reason
A planner would release the order once readiness passes, or record an approved exception. No release against missing critical components without an approved shortage reason.
ProductionOrder
- 04execute
Confirm on the floor
Operators would confirm operation start, stop, and partial completion; WIP moves, labor and machine time, yield, scrap, rework, and downtime captured as evidence.
OperationConfirmation
- 05quality
Hold or release
Inspection checkpoints would record measurements and either release, hold, or route to rework. A failed mandatory checkpoint would block completion until approved.
QualityCheckpoint
- 06complete
Declare and hand off
Finished output would be declared with lot or serial traceability, handed to WMS for finished-goods receipt, and costing evidence handed to FMS. Close would need both.
OutputDeclaration
designed production-order lifecycle
The dashed state is the designed exception branch - work could be held for quality, material, safety, maintenance, or engineering reasons and resumed by the owning function. Cancelled, scrapped, split, merged, and reworked are the other designed states. All designed, none built.
The proposed Arin-over-MCP surface
The blueprint designs MFG as agent-ready. Arin would read production facts over MCP - readiness, timelines, quality holds, shortage and variance evidence - explain the exception, and prepare an approval-bound action. It mirrors the family's proven Arin-drives-SCM po.approve path behind a human gate. Arin would never release, substitute, close, or override directly; MFG would validate policy and human approval, then execute.
mfg:schedule:write
- production-order.plan
- production-order.release
mfg:operation:write
- production-order.hold
- production-order.resume
- operation.confirm
mfg:material:write
- material.substitution.request
mfg:quality:write
- quality.hold.release
- output.declare
- production-order.close
Every mutating tool would declare its required mfg:* scope, carry an idempotency key, and high-impact actions would require an approval state and a reason. Reads and writes would share one service layer behind REST, the UI, and MCP - the discipline WMS and SCM already run in the family. This surface is designed, not built.
The designed family boundary
MFG is designed to own the production record and nothing else. It would not replace WMS, SCM, FMS, EAM, PLM, an external MES, or Arin. The blueprint draws a strict boundary: no shared database tables, integration only over REST, MCP, and events, and fail-closed behaviour when another system's evidence is missing.
ICE
Identity and tokens for the family.
MFG would be an ICE resource server, validating ICE-issued tokens on mfg:* scopes.
Arin
Agent and decision-support layer.
Would read MFG facts and prepare approval-bound actions over MFG MCP - never own truth.
SCM
Procurement, suppliers, purchase orders.
Would supply incoming-supply and delay evidence; MFG owns the production effect.
WMS
Physical stock and locations.
Would reserve, stage, issue materials and receive finished goods; MFG owns production use.
EAM
Asset registry and maintenance.
Would supply availability and downtime evidence; MFG owns the production impact.
RMS / FMS / PLM
Demand, finance, engineering truth.
Demand signals in, costing evidence out, engineering BOM in - all through contracts.
the designed boundary
Arin would explain, classify, recommend, and prepare actions. MFG would validate the action, enforce approval and shop-floor controls, and execute through its own services. This is the intended arrangement, not running software.
Blueprint - in design
Help shape the production record.
MFG is a blueprint, not a product - there is nothing to download and no sign-up. If turning demand, materials, capacity, and shop-floor evidence into one controlled, agent-ready production record is a problem you have, tell us about your manufacturing stack. Early conversations shape which phase gets built first.