What is this application worth?

POST

/v1/score

curl --request POST \
  --url https://stitch-collector-dev.magic-blocks.workers.dev/v1/score \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{ "loan_amount": 250000, "property_value": 357143, "credit_range": "Above 740", "loan_purpose": "Refi - CO", "product_grouping": "DSCR" }'

• Development. Production puts the collector on a first-party host per brand (t.makeabeeline.com, t.beelineequitynow.com); nothing in these contracts changes, only the hostname.

For whoever holds the application’s figures at the point its sections are complete — today, most likely the application’s own backend. Same body and answer as the EV Worker’s /score, in dollars and in cents. Logs nothing: a score without a loan is a what-if; the predictions that count are the ones written against loans through /v1/conversion. Do not call it again with updated figures — the value is corrected afterwards only by Stitch’s own restatements.

Authorizations

• webhookKey

Request Body ^required

Media type application/json

The five fields the borrower entered, as typed. Unknown labels do not fail the request: the field falls back to its reference level and a warning is returned.

object

loan_amount

required

Dollars. The only field whose absence is a 400.

number

property_value

Dollars, as the applicant typed it (purchase price for purchases). The round-number signal lives here, so never send a value derived from a ratio.

number

credit_range

One of:

string

string

number

number

fico

The pulled score, when present; the FICO model is then used.

integer

loan_purpose

Purchase, Refi - CO, Refi - RT; aliases accepted (cash-out refinance, rate/term).

string

product_grouping

DSCR, Conv, Bank Stmt/Non-QM, Govt/Jumbo/Other; aliases accepted (conventional, bank statement, FHA).

string

occupancy

Accepted; not used by the shipped models.

string

Responses

200

The prediction.

Media type application/json

The same answer the EV Worker gave, in dollars and in cents. Report conversion_value as the value of the application conversion you report to Google.

object

prediction_id

string format: uuid

ev

Expected value in dollars: p_close × rev_if_close, after calibration.

number

conversion_value

What to report to the ad platform. Equal to ev.

number

p_close

number

<= 1

rev_if_close

number

model_used

string

Allowed values: credit_range fico

model

The model file’s name and version.

string

trained_on

string | null format: date

train_window

Array<string> | null

>= 2 items <= 2 items

calibration_multiplier

number

calibration

object

manual

number

auto

number

auto_source

string

Allowed values: off

inputs_used

object

key

additional properties

any

warnings

Array<string>

complete

Whether every recommended field was present.

boolean

ev_cents

integer

conversion_value_cents

integer

rev_if_close_cents

integer

Example

{
  "model_used": "credit_range",
  "calibration": {
    "auto_source": "off"
  }
}

400

Malformed JSON or a payload missing what the endpoint needs. The body says which.

Media type application/json

object

error

required

What was wrong, in a sentence. A 401 carries unauthorized and nothing more.

string

Example ^generated

{
  "error": "example"
}

401

No key, or a key the collector does not know. Nothing was written.

Media type application/json

object

error

required

What was wrong, in a sentence. A 401 carries unauthorized and nothing more.

string

Example

{
  "error": "unauthorized"
}

405

Only the method shown is served.