Reverse Engineering Your Software program Structure with Claude Code to Assist Claude Code – O’Reilly

I’ve been utilizing Claude Code for quite a lot of functions, and one factor I’ve realized is that the extra it understands concerning the performance of the system (the area, the use circumstances, the end-to-end flows), the extra it will possibly assist me.

For instance, after I paste a manufacturing error log, Claude can learn the stack hint, determine the affected code, and inform me if there’s a bug. However when the difficulty is extra advanced, like a buyer assist ticket, and there’s no stack hint, Claude is much less helpful.

The principle problem is that end-to-end processes are lengthy and sophisticated, spanning many code repositories. So simply asking Claude Code to research a single repository wasn’t going to work (and the default /init wasn’t producing adequate element even for this single codebase).

So I made a decision to make use of Claude Code to research the system to map out the end-to-end flows related to the area I work in in order that Claude Code (and people) can use this to deal with extra advanced challenges.

This publish shares what I knocked collectively in in the future, constructing on data and tooling I’ve already gained from actual work examples and experiments.

That is one publish in a sequence. Yow will discover the opposite posts right here.

This publish was written 100% by me. I requested Claude to generate the anonymized instance on the finish mirroring the kind of content material and magnificence utilized in the actual examples I created.

Setting the Preliminary Context

To start my undertaking, I created a really mild necessities doc:

# AI Structure Evaluation

This doc accommodates the directions for an necessary job - utilizing AI to outline the structure of this method, in order that it may be utilized by people and AI brokers to extra simply perceive the system.

## Goal

Map out all the flows that this utility is concerned in (use sub brokers the place essential to work in parallel). A circulation ought to map out the end-to-process from an motion within the UI (within the [readacted] repo) to a BFF, to backend APIs, or flows which might be triggered by occasions.

Flows ought to be documented in Mermaid format to permit AI brokers to grasp, for versioning (in git), and for simple visualization.

## Necessities

Every circulation ought to have a descriptive identify and will embrace:

1. The URL path of the web page the place the interplay is triggered
2. The URL path of the BFF endpoint (and the repository it lives in)
3. The URL path of calls made to downstream providers
4. Any database interactions
5. Any occasions produced or consumed (full identify of occasion e.g. orders.orderPlaced)
6. Shoppers of occasions (if straightforward to determine)
7. Any workflows triggered (just like the synchronizeOrder)

To do that, you will want to look in different repositories, which will be discovered within the guardian folder. The GitHub consumer may also be used if mandatory.

The checklist of flows ought to stay in ../flows/index.md and every particular person circulation ought to be outlined in a separate folder.

# The place to seek out data

- /docs/structure accommodates varied folders describing the design of this method and area data
- Every API undertaking on this repository ([redacted], [redacted]) has an openapi.json. This have to be used to determine all flows and validate. The [redacted] and [redacted] repositories even have openapi spec information
- The entities within the area [redacted], [redacted] [redacted] have methodology that clearly describe the area operations that may be carried out on them. Equally, every operation is invoke from a use case that clearly describes the use case

The output I need is end-to-end flows like:
UI -> BFF -> API -> replace DB -> publish occasion -> handler -> use case -> publish occasion -> …

I don’t need 10 totally different sorts of structure diagrams and totally different ranges of element. I need Claude Code to grasp the conduct of the system so it will possibly determine anomalies (by manufacturing information and logs) and analyze the influence of potential modifications.

I additionally created some mild details about the system in these two information:

The area ideas file explains the entities within the system. Very temporary rationalization. The system overview file explains the connection between this codebase and different repositories, which is essential. Once more, it’s very mild—a bullet checklist of repository names and one or two sentences describing their relationship to this one.

Looking out throughout a number of repositories

The directions for this job stay inside the primary repository of the area I work in. That is the middle of the universe for this agent, nevertheless it wants to have the ability to learn different repositories to hitch up the end-to-end circulation.

The answer I take advantage of for that is within the description above:

To do that, you will want to look in different repositories which will be discovered within the guardian folder. The GitHub consumer may also be used if mandatory.

I give Claude the next permissions in .claude/settings.native.json and it will possibly then entry all of the repositories on machine or use the GitHub consumer if it thinks there are repositories I don’t have accessible regionally:

"permissions": {
   "permit": [
     ...
     "Read(//Users/nicktune/code/**)",
     ...

Telling Claude where to look

You’ll notice the requirements also give Claude tips on where to look for key information like Open API spec files, which are like an index of the operations the application supports.

This is useful as a validation mechanism later in the flow. I would ask Claude, “List all of the API endpoints and events produced or consumed by this application—are there any that aren’t part of any flows.” I can then see if we may have missed anything important.

Mapping the First Flow

I put Claude in plan mode and asked it to read the file. It then popped up a short questionnaire asking me about my needs and preferences. One of the questions it asked was process related: Should we map out the whole system in parallel, work step-by-step, etc.?

I said, let’s do the first one together and use this as a template for the others to follow.

It took about two hours to build the first flow as I reviewed what Claude produced and gave feedback on what I needed. For example, at first it created a sequence diagram which looked nice but was too hard to read for complex flows that involve many repositories.

Eventually, we settled on horizontal flow diagrams where each repository is a container and we defined what steps could be. At first, it went too granular with the steps adding individual method calls.

### diagram.mermaid Requirements
**CRITICAL RULES:**
1. **File Format**: Must be pure Mermaid syntax with `.mermaid` extension- NO markdown headers
 - NO markdown code fences (no ` ```mermaid ` wrapper)
 - Starts directly with `flowchart LR`
2. **Use Swimlane Format**: `flowchart LR` (left-to-right with horizontal swimlanes)
 - Each repository is a horizontal swimlane (subgraph)
 - Flow progresses left to right
 - Swimlane labels should be prominent (use emoji for visibility)
 - Example: `subgraph systemA["🔧 systemA"]`
3. **Programs as Containers**:
 - Every repository MUST be a `subgraph` (horizontal swimlane)
 - Repository identify is the subgraph label
 - Operations are nodes contained in the subgraph
 - Use `course LR` inside every subgraph
4. **Legitimate Step Sorts** - A step within the diagram can ONLY be one of many following:
 - **HTTP Endpoint**: Full endpoint path (e.g., `POST /blah/{blahId}/subblah`)
 - **Combination Methodology Name**: Area methodology on an combination (e.g., `Order.place`, `Transport.organiz`)
 - **Database Operation**: Proven with cylinder form `[(Database: INSERT order)]`
 - **Occasion Publication**: (e.g., `Publish: personal.ordering.order.positioned`)
 - **Workflow Set off**: Should be labeled as workflow (e.g., `⚙️ Workflow: syncOrders`)
 - **Workflow Step**: Any step inside a workflow MUST embrace the workflow identify as prefix (e.g., `syncOrderWorkflow: Replace legacy order`, `updateOrderInfo: POST /legacy/fill-order`)
 - **Lambda Invocation**: (e.g., `Lambda: blah-blah-lambda-blah-blah`)
 - **UI Actions**: Person interactions (e.g., `Present modal type`, `Person enters firstName, lastName`)

I’ve added an anonymized circulation on the finish of this doc.

I additionally had so as to add some corrections to Claude to make sure it was wanting in all the suitable locations and understanding what sure ideas imply in different components of our system; we weren’t simply iterating on the diagram for 2 hours.

Selecting the Subsequent Stream

After the primary circulation, the subsequent flows went a lot quicker. I verified the output of every circulation and gave Claude suggestions, however usually round quarter-hour, and more often than not it was working so I may do different issues whereas ready.

One of many fascinating challenges was deciding which flows are literally wanted? What’s a circulation? The place ought to a circulation begin and finish? What about relationships between flows?

Right here I used to be within the driving seat fairly a bit. I requested Claude to suggest flows (simply checklist them earlier than analyzing) after which I requested it to indicate me how every API endpoint and occasion match into the flows, and we used that to iterate a bit.

One of many issues I needed to do after Claude had produced the primary draft is to ask, “Are you certain there aren’t any different customers for these occasions that aren’t listed right here?” It will then do a extra thorough search and generally discover customers in repositories I didn’t have regionally. (It will use GitHub search.)

Studying worth

As I reviewed every use case, I used to be studying issues concerning the system that I didn’t totally perceive or possibly there have been nuances I wasn’t conscious of. This alone would have justified all the hassle I spent on this.

Then I began to think about the worth for people who find themselves new to a codebase or a legacy system that no one understands anymore. Or possibly somebody who works in a special crew and wishes to determine how a bug or a change of their area is expounded to different domains.

Evolving the Necessities

As we went via the method, I frequently advised Claude to replace the necessities file. So after we’d completed the primary circulation we had directions like this added to the file:

## Documentation Construction

Every circulation have to be documented in a separate folder with the next construction:

```
docs/structure/flows/[flow-name]/
├── README.md              # Full documentation (all content material in a single file)
└── diagram.mermaid        # Mermaid diagram
```

**IMPORTANT**: Use ONLY these two information. Do NOT create separate diagram-notes.md or different information. Hold all documentation consolidated in README.md for simpler upkeep.

### README.md Contents

**Use the blueprint as a template**: `docs/structure/flows/[redacted]/`

The file is now 449 traces lengthy.

One of many causes I did this was in order that I may begin a brand new Claude session, now or sooner or later, with no fully clear context window and execute the method to get related outcomes.

I did really use a brand new session to map every new circulation to validate that the necessities have been considerably repeatable. On the whole they have been, however typically Claude would ignore some components of the necessities. So on the finish, I advised it to overview the necessities and evaluate the outputs, and it will often determine many of the errors it had made and repair them.

Right here’s an instance of a few of the guidelines that began to construct up. Some have been to make sure Claude produced the suitable sort of output, and a few have been to assist Claude keep away from widespread errors like Mermaid syntax errors.

### 2. Hint Workflows to Their Closing Occasion
**Drawback**: Lacking occasions since you do not learn the precise workflow implementation.

**Rule**: Whenever you encounter a workflow, you MUST:
1. Discover the workflow definition file (often `.asl.json` for AWS Step Features)
2. Learn the ENTIRE workflow to see ALL occasions it publishes
3. Doc EVERY occasion in sequence

**Instance from our blueprint**:
- We initially thought `[readacted]` ended with `[readacted]`
- Studying `[redacted].asl.json` revealed it really ends with `[readacted]`
- This occasion was CRITICAL to the circulation persevering with

**File areas**:
- Integrations workflows: `[another-repo]/domains/*/workflows/*.asl.json`
- Search for `SendEvent` or `publish` actions within the workflow steps

Claude Code by no means follows each instruction as described. However including these guidelines does appear to extend the standard and scale back the variety of iterations wanted.

Testing the Output

After I mapped out 4 or 5 of the flows, I made a decision it was time to check it out—Does this data really assist Claude Code, or is it primarily only a human viewers that advantages from the outputs?

So I went into our assist ticket system and picked a ticket that appeared advanced. I requested my investigation agent to take a look at the ticket and determine the issue. However this time I added an additional step into its directions:

4. Determine the affected flows in /docs/structure/flows

As a part of its evaluation, it mentioned:

Let me discover the structure flows to grasp how [redacted] and [redacted] ought to be dealt with:

Then it produced its evaluation appropriately figuring out the circulation and anticipated behaviors:

## Affected Stream

**[BlahA Upload and BlahB History Thing](../structure/flows/[redacted]/README.md)**

The [redacted] workflow ought to populate [redacted] information through:
- [redacted] choice ([redacted] commonplace OR [redacted] enhanced primarily based on `[redacted]` flag)
- Parallel execution of: DomainA, DomainB, DomainC, DomainD
- Publishing `order.blahBlahed` occasion on completion
…

And for the subsequent steps, it wished to question the occasions revealed by the system to confirm that what really occurred matched the anticipated behaviors outlined within the circulation:

### Step 1: Question [redacted] Occasions for Each [redacted]
**Goal:** Evaluate occasion historical past to determine what occurred otherwise between outdated and new [redacted]

**What to search for:**
- `[redacted event name]` - Did [redacted] full for brand new [redacted]?
- `[redacted event name]` - Have been [redacted] created?
- `[redacted event name]` - Have been [redacted] created?
- `[redacted event name]` - Did workflow full for brand new [redacted]?
- `[redacted event name]` - Closing affirmation occasion
- Any error/failure occasions associated to the workflow

Beforehand, Claude would have needed to analyze the codebase to work out what ought to have occurred. It takes a very long time and takes up a number of context window for advanced duties, and the evaluation must be verified.

Now, Claude is aware of instantly concerning the particular workflow and affected behaviors and may instantly start planning an investigation (if the documentation is correct sufficient). This evaluation is structured with the important thing data that I must see. I don’t must iterate with Claude to provide an evaluation within the format I want.

On this case, Claude didn’t resolve the issue instantly, however the dialog was extra like I may need with a crew member—somebody who has a deeper understanding of how the system works and what is perhaps mistaken right here moderately than simply utilizing Claude to research patterns in information, learn stack traces, or summarize textual content descriptions of the issue.

Accuracy and Hallucinations

I do assume it’s proper to be involved about accuracy. We don’t need to make necessary decisions about our system primarily based on incomplete or incorrect particulars. And there have been important inaccuracies that I needed to spot and proper. (Think about if I didn’t know they have been mistaken.)

I explored the problem of accuracy in this later publishdisplaying how we will use deterministic instruments like ts-morph to construct the mannequin that people and AI can each profit from.

So right here’s what I’m considering:

1. Typically we don’t want excellent accuracy. So long as the agent picks the suitable path, it will possibly reinspect sure particulars or dive deeper as wanted.
2. We are able to construct checks and steps in into our CI pipelines to replace issues.
3. Frequently destroy and regenerate the flows (as soon as 1 / 4?).
4. Construct verification brokers or swarms.

After I noticed an error and requested a brand new agent to research the circulation for inaccuracies, it rescanned the code and located what I noticed. So I believe possibility 4 may be very credible—it’s simply extra effort to construct a verification system (which may make the general effort not value it).

However I’m unsure that is the optimum means of approaching the scenario. As an alternative…

The Subsequent Section of Platform Engineering

Avoiding the necessity to reverse engineer these flows might be key. And I’m beginning to assume it will change into the primary problem for platform engineering groups: How can we construct frameworks and tooling that expose our system as a graph of dependencies? Constructed into our platform in order that AI brokers don’t must reverse engineer; they’ll simply seek the advice of the supply of fact.

Issues ought to all occur transparently for software program engineers—you comply with the platform paved path, and every part simply works. Corporations that do that, and particularly startups with no legacy, may immensely revenue from AI brokers.

Instruments like EventCatalog are in a powerful place right here.

Instance Stream

I simply requested Claude to translate considered one of my firm’s area flows right into a boring ecommerce instance. The design and naming shouldn’t be necessary; the kind of data and the visualization is what I’m making an attempt to convey.

Bear in mind, that is primarily based on in the future of hacking round. I’m certain there are many enchancment alternatives right here. Let me know you probably have seen something higher.

The README

# Place Order with Fee and Achievement

**Standing**: Lively
**Sort**: Write Operation
**Complexity**: Excessive
**Final Up to date**: 2025-10-19

## Overview

This circulation paperwork the method of putting an order in an ecommerce system, together with cost authorization, stock reservation, and cargo creation. That is the baseline order placement expertise the place:
- Orders begin with `standing: 'pending'`
- Fee is permitted earlier than stock reservation
- Stock is reserved upon profitable cost
- Cargo is created after stock reservation

## Stream Boundaries

**Begin**: Buyer clicks "Place Order" button on checkout web page

**Finish**: Publication of `transport.shipment-created` occasion (public occasion with `DOMAIN` scope)

**Scope**: This circulation covers your complete course of from preliminary order submission via cost authorization, stock reservation, cargo creation, and all asynchronous negative effects triggered by these operations.

## Fast Reference

### API Endpoints

| Endpoint | Methodology | Repository | Goal |
|----------|--------|------------|---------|
| `/checkout` | GET | storefront-app | Checkout web page |
| `/api/orders` | POST | order-api | Creates order |
| `/api/funds/authorize` | POST | payment-api | Authorizes cost |
| `/api/stock/reserve` | POST | inventory-api | Reserves stock |
| `/api/shipments` | POST | shipping-api | Creates cargo |
| `/api/orders/{orderId}/standing` | GET | order-api | Frontend polls for order standing |

### Occasions Reference

| Occasion Identify | Area | Topic | Goal | Shoppers |
|------------|--------|---------|---------|-----------|
| `personal.orders.order.created` | ORDERS | order | Order creation | PaymentHandler, AnalyticsHandler |
| `personal.funds.cost.licensed` | PAYMENTS | cost | Fee licensed | InventoryReservationHandler |
| `personal.funds.cost.failed` | PAYMENTS | cost | Fee failed | OrderCancellationHandler |
| `personal.stock.inventory.reserved` | INVENTORY | inventory | Stock reserved | ShipmentCreationHandler |
| `personal.stock.inventory.inadequate` | INVENTORY | inventory | Inadequate inventory | OrderCancellationHandler |
| `personal.transport.cargo.created` | SHIPPING | cargo | Cargo created | NotificationHandler |
| `transport.shipment-created` | SHIPPING | cargo | **PUBLIC** Cargo created | Exterior customers |

### Database Tables

| Desk | Operation | Key Fields | Goal |
|-------|-----------|------------|---------|
| `orders` | INSERT | orderId, customerId, standing="pending", totalAmount | Order combination storage |
| `order_items` | INSERT | orderItemId, orderId, productId, amount, value | Order line gadgets |
| `funds` | INSERT | paymentId, orderId, quantity, standing="licensed" | Fee combination storage |
| `inventory_reservations` | INSERT | reservationId, orderId, productId, amount | Stock reservation monitoring |
| `shipments` | INSERT | shipmentId, orderId, trackingNumber, standing="pending" | Cargo combination storage |

### Area Operations

| Combination | Methodology | Goal |
|-----------|--------|---------|
| Order | `Order.create()` | Creates new order with pending standing |
| Order | `Order.confirmPayment()` | Marks cost as confirmed |
| Order | `Order.cancel()` | Cancels order as a consequence of cost or stock failure |
| Fee | `Fee.authorize()` | Authorizes cost for order |
| Fee | `Fee.seize()` | Captures licensed cost |
| Stock | `Stock.reserve()` | Reserves inventory for order |
| Cargo | `Cargo.create()` | Creates cargo for order |

## Key Traits

| Side | Worth |
|--------|-------|
| Order Standing | Makes use of `standing` subject: `'pending'` → `'confirmed'` → `'shipped'` |
| Fee Standing | Makes use of `standing` subject: `'pending'` → `'licensed'` → `'captured'` |
| Stock Technique | Reserve-on-payment strategy |
| Cargo Standing | Makes use of `standing` subject: `'pending'` → `'prepared'` → `'shipped'` |

## Stream Steps

1. **Buyer** navigates to checkout web page in storefront-app (`/checkout`)
2. **Buyer** evaluations order particulars and clicks "Place Order" button
3. **storefront-app UI** reveals loading state with order affirmation message
4. **storefront-app** sends POST request to order-api (`/api/orders`)
  - Request contains: customerId, gadgets (productId, amount, value), shippingAddress, billingAddress
5. **order-api** creates Order combination with `standing: 'pending'` and persists to database
6. **order-api** creates OrderItem data for every merchandise within the order
7. **order-api** publishes `personal.orders.order.created` occasion
8. **order-api** returns orderId and order particulars to storefront-app
9. **storefront-app** redirects buyer to order affirmation web page

### Asynchronous Facet Results - Fee Processing

10. **order-events-consumer** receives `personal.orders.order.created` occasion
11. **PaymentHandler** processes the occasion
   - Calls payment-api to authorize cost
12. **payment-api** calls exterior cost gateway (Stripe, PayPal, and so on.)
13. **payment-api** creates Fee combination with `standing: 'licensed'` and persists to database
14. **payment-api** publishes `personal.funds.cost.licensed` occasion (on success)
   - OR publishes `personal.funds.cost.failed` occasion (on failure)

### Asynchronous Facet Results - Stock Reservation

15. **payment-events-consumer** receives `personal.funds.cost.licensed` occasion
16. **InventoryReservationHandler** processes the occasion
   - Calls inventory-api to order inventory
17. **inventory-api** masses Stock combination for every product
18. **inventory-api** calls `Stock.reserve()` for every order merchandise
   - Validates adequate inventory accessible
   - Creates reservation document
   - Decrements accessible inventory
19. **inventory-api** creates InventoryReservation data and persists to database
20. **inventory-api** publishes `personal.stock.inventory.reserved` occasion (on success)
   - OR publishes `personal.stock.inventory.inadequate` occasion (on failure)

### Asynchronous Facet Results - Cargo Creation

21. **inventory-events-consumer** receives `personal.stock.inventory.reserved` occasion
22. **ShipmentCreationHandler** processes the occasion
   - Calls shipping-api to create cargo
23. **shipping-api** creates Cargo combination with `standing: 'pending'` and persists to database
24. **shipping-api** calls exterior transport service API to generate monitoring quantity
25. **shipping-api** updates Cargo with trackingNumber
26. **shipping-api** publishes `personal.transport.cargo.created` occasion
27. **shipping-events-consumer** receives `personal.transport.cargo.created` occasion
   - **ShipmentCreatedPublicHandler** processes the occasion
   - Hundreds cargo from repository to get full cargo particulars
   - Publishes public occasion: `transport.shipment-created`
   - **This marks the END of the circulation**

### Order Standing Updates

28. All through the circulation, order-api receives occasions and updates order standing:
   - On `personal.funds.cost.licensed`: Updates order with paymentId
   - On `personal.stock.inventory.reserved`: Updates order to `standing: 'confirmed'`
   - On `personal.transport.cargo.created`: Updates order to `standing: 'shipped'`

### Failure Eventualities

**Fee Failure**:
- On `personal.funds.cost.failed`: OrderCancellationHandler cancels order
- Order standing up to date to `'cancelled'`
- Buyer notified through e-mail

**Stock Failure**:
- On `personal.stock.inventory.inadequate`: OrderCancellationHandler cancels order
- Fee authorization is voided
- Order standing up to date to `'cancelled'`
- Buyer notified through e-mail with choice to backorder

## Repositories Concerned

- **storefront-app**: Frontend UI
- **order-api**: Order area
- **payment-api**: Fee area
- **inventory-api**: Stock area
- **shipping-api**: Transport and success area
- **notification-api**: Buyer notifications

## Associated Flows

- **Course of Refund**: Stream for dealing with order refunds and returns
- **Replace Cargo Standing**: Stream for monitoring cargo supply standing
- **Stock Reconciliation**: Stream for syncing stock counts with warehouse programs

## Occasions Produced

| Occasion | Goal |
|-------|---------|
| `personal.orders.order.created` | Notifies {that a} new order has been created |
| `personal.funds.cost.licensed` | Notifies that cost has been licensed |
| `personal.funds.cost.failed` | Notifies that cost authorization failed |
| `personal.stock.inventory.reserved` | Notifies that stock has been reserved |
| `personal.stock.inventory.inadequate` | Notifies that inadequate stock is on the market |
| `personal.transport.cargo.created` | Inside occasion that cargo has been created |
| `transport.shipment-created` | **Public occasion** that cargo is created and prepared for service pickup |

## Occasion Shoppers

### `personal.orders.order.created` Shoppers

#### 1. order-events-consumer

**Handler**: `PaymentHandler`

**Goal**: Initiates cost authorization course of

**Actions**:
- Subscribes to occasion
- Calls `AuthorizePayment` use case
- Invokes payment-api to authorize cost with cost gateway
- Publishes cost outcome occasion

#### 2. order-events-consumer

**Handler**: `AnalyticsHandler`

**Goal**: Tracks order creation for analytics

**Actions**:
- Subscribes to occasion
- Sends order information to analytics platform
- Updates conversion monitoring

### `personal.funds.cost.licensed` Client

#### payment-events-consumer

**Handler**: `InventoryReservationHandler`

**Goal**: Reserves stock after profitable cost

**Actions**:
- Subscribes to occasion
- Calls `ReserveInventory` use case
- Hundreds order particulars
- Calls inventory-api to order inventory for every merchandise
- Publishes stock reservation outcome occasion

### `personal.funds.cost.failed` Client

#### payment-events-consumer

**Handler**: `OrderCancellationHandler`

**Goal**: Cancels order when cost fails

**Actions**:
- Subscribes to occasion
- Calls `CancelOrder` use case
- Updates order standing to 'cancelled'
- Triggers buyer notification

### `personal.stock.inventory.reserved` Client

#### inventory-events-consumer

**Handler**: `ShipmentCreationHandler`

**Goal**: Creates cargo after stock reservation

**Actions**:
- Subscribes to occasion
- Calls `CreateShipment` use case
- Calls shipping-api to create cargo document
- Integrates with transport service API for monitoring quantity
- Publishes cargo created occasion

### `personal.stock.inventory.inadequate` Client

#### inventory-events-consumer

**Handler**: `OrderCancellationHandler`

**Goal**: Cancels order when stock is inadequate

**Actions**:
- Subscribes to occasion
- Calls `CancelOrder` use case
- Voids cost authorization
- Updates order standing to 'cancelled'
- Triggers buyer notification with backorder possibility

### `personal.transport.cargo.created` Client

#### shipping-events-consumer

**Handler**: `ShipmentCreatedPublicHandler`

**Goal**: Converts personal cargo occasion to public occasion

**Actions**:
- Subscribes to `personal.transport.cargo.created` occasion
- Hundreds cargo from repository
- Publishes public occasion: `transport.shipment-created`

**Handler**: `NotificationHandler`

**Goal**: Notifies buyer of cargo creation

**Actions**:
- Subscribes to occasion
- Sends affirmation e-mail with monitoring quantity
- Sends SMS notification (if opted in)

## Database Operations

### orders Desk
- **Operation**: INSERT (through upsert)
- **Key Fields**: orderId, customerId, standing="pending", totalAmount, createdAt
- **Repository**: `OrderRepository`

### order_items Desk
- **Operation**: INSERT (batch)
- **Key Fields**: orderItemId, orderId, productId, amount, value
- **Repository**: `OrderItemRepository`

### funds Desk
- **Operation**: INSERT (through upsert)
- **Key Fields**: paymentId, orderId, quantity, standing="licensed", gatewayTransactionId
- **Repository**: `PaymentRepository`

### inventory_reservations Desk
- **Operation**: INSERT (through upsert)
- **Key Fields**: reservationId, orderId, productId, amount, reservedAt
- **Repository**: `InventoryReservationRepository`

### shipments Desk
- **Operation**: INSERT (through upsert)
- **Key Fields**: shipmentId, orderId, trackingNumber, standing="pending", service
- **Repository**: `ShipmentRepository`

## Exterior Integrations

- **Fee Gateway Integration**: Authorizes and captures funds through Stripe API
 - Endpoint: `/v1/payment_intents`
 - Synchronous name throughout cost authorization

- **Transport Provider Integration**: Generates monitoring numbers through service API
 - Endpoint: `/api/v1/shipments`
 - Synchronous name throughout cargo creation

## What Occurs After This Stream

This circulation ends with the publication of the `transport.shipment-created` public occasion, which marks the order as totally processed and prepared for service pickup.

### State at Stream Completion
- Order: `standing: 'shipped'`
- Fee: `standing: 'licensed'` (might be captured on precise cargo)
- Stock: Inventory reserved and decremented
- Cargo: `standing: 'pending'`, trackingNumber assigned

### Subsequent Steps
After this circulation completes:
- Warehouse crew picks and packs the order
- Provider picks up the cargo
- Transport standing updates circulation tracks supply
- Fee is captured upon confirmed cargo
- Buyer can observe order through monitoring quantity

### Exterior System Integration
As soon as the `transport.shipment-created` occasion is revealed:
- Warehouse administration system begins decide/pack course of
- Buyer notification system sends monitoring updates
- Logistics companions obtain cargo manifest
- Analytics programs observe success metrics

## Diagram

See [diagram.mermaid](./diagram.mermaid) for the whole visible circulation displaying the development via programs with horizontal swim lanes for every service.

The Mermaid:

flowchart LR
   Begin([Customer clicks Place Order
on checkout page])

   subgraph storefront["🌐 storefront-app"]
       course LR
       ShowCheckout[Show checkout page]
       CustomerReview[Customer reviews order]
       ShowConfirmation[Show order
confirmation page]
   finish

 CustomerWaitsForShipment([Customer receives
shipment notification])

subgraph orderService["📦 order-api"]
       course LR
       CreateOrderEndpoint["POST /api/orders"]
       OrderCreate[Order.create]
       OrderDB[(Database:
INSERT orders,
order_items)]
       PublishOrderCreated["Publish: private.orders
.order.created"]
       ReceivePaymentAuth["Receive: private.payments
.payment.authorized"]
       UpdateOrderPayment[Update order
with paymentId]
       ReceiveStockReserved["Receive: private.inventory
.stock.reserved"]
       OrderConfirm[Order.confirmPayment]
       UpdateOrderConfirmed[(Database:
UPDATE orders
status="confirmed")]
       ReceiveShipmentCreated["Receive: private.shipping
.shipment.created"]
       UpdateOrderShipped[(Database:
UPDATE orders
status="shipped")]
   finish

 subgraph paymentService["💳 payment-api"]
       course LR
       ReceiveOrderCreated["Receive: private.orders
.order.created"]
       AuthorizeEndpoint["POST /api/payments/
authorize"]
       PaymentGateway["External: Payment
Gateway API
(Stripe)"]
       PaymentAuthorize[Payment.authorize]
       PaymentDB[(Database:
INSERT payments)]
       PublishPaymentAuth["Publish: private.payments
.payment.authorized"]
   finish

 subgraph inventoryService["📊 inventory-api"]
       course LR
       ReceivePaymentAuth2["Receive: private.payments
.payment.authorized"]
       ReserveEndpoint["POST /api/inventory/
reserve"]
       InventoryReserve[Inventory.reserve]
       InventoryDB[(Database:
INSERT inventory_reservations
UPDATE product stock)]
       PublishStockReserved["Publish: private.inventory
.stock.reserved"]
   finish

 subgraph shippingService["🚚 shipping-api"]
       course LR
       ReceiveStockReserved2["Receive: private.inventory
.stock.reserved"]
       CreateShipmentEndpoint["POST /api/shipments"]
       CarrierAPI["External: Shipping
Carrier API
(FedEx/UPS)"]
       ShipmentCreate[Shipment.create]
       ShipmentDB[(Database:
INSERT shipments)]
       PublishShipmentCreated["Publish: private.shipping
.shipment.created"]
       ReceiveShipmentCreatedPrivate["Receive: private.shipping
.shipment.created"]
       LoadShipment[Load shipment
from repository]
       PublishPublicEvent["Publish: shipping
.shipment-created"]
       FlowEnd([Flow End:
Public event published])
   finish

   Begin --> ShowCheckout
   ShowCheckout --> CustomerReview
   CustomerReview --> CreateOrderEndpoint
   CreateOrderEndpoint --> OrderCreate
   OrderCreate --> OrderDB
   OrderDB --> PublishOrderCreated
   PublishOrderCreated --> ShowConfirmation

   PublishOrderCreated -.-> ReceiveOrderCreated
   ReceiveOrderCreated --> AuthorizeEndpoint
   AuthorizeEndpoint --> PaymentGateway
   PaymentGateway --> PaymentAuthorize
   PaymentAuthorize --> PaymentDB
   PaymentDB --> PublishPaymentAuth

   PublishPaymentAuth -.-> ReceivePaymentAuth
   ReceivePaymentAuth --> UpdateOrderPayment

   PublishPaymentAuth -.-> ReceivePaymentAuth2
   ReceivePaymentAuth2 --> ReserveEndpoint
   ReserveEndpoint --> InventoryReserve
   InventoryReserve --> InventoryDB
   InventoryDB --> PublishStockReserved

   PublishStockReserved -.-> ReceiveStockReserved
   ReceiveStockReserved --> OrderConfirm
   OrderConfirm --> UpdateOrderConfirmed

   PublishStockReserved -.-> ReceiveStockReserved2
   ReceiveStockReserved2 --> CreateShipmentEndpoint
   CreateShipmentEndpoint --> CarrierAPI
   CarrierAPI --> ShipmentCreate
   ShipmentCreate --> ShipmentDB
   ShipmentDB --> PublishShipmentCreated

   PublishShipmentCreated -.-> ReceiveShipmentCreated
   ReceiveShipmentCreated --> UpdateOrderShipped

   PublishShipmentCreated -.-> ReceiveShipmentCreatedPrivate
   ReceiveShipmentCreatedPrivate --> LoadShipment
   LoadShipment --> PublishPublicEvent
   PublishPublicEvent --> FlowEnd

   FlowEnd -.-> CustomerWaitsForShipment

   model Begin fill:#e1f5e1
   model FlowEnd fill:#ffe1e1
   model CustomerWaitsForShipment fill:#e1f5e1
   model PublishOrderCreated fill:#fff4e1
   model PublishPaymentAuth fill:#fff4e1
   model PublishStockReserved fill:#fff4e1
   model PublishShipmentCreated fill:#fff4e1
   model PublishPublicEvent fill:#fff4e1
   model OrderDB fill:#e1f0ff
   model PaymentDB fill:#e1f0ff
   model InventoryDB fill:#e1f0ff
   model ShipmentDB fill:#e1f0ff
   model UpdateOrderConfirmed fill:#e1f0ff
   model UpdateOrderShipped fill:#e1f0ff
   model PaymentGateway fill:#ffe1f5
   model CarrierAPI fill:#ffe1f5