A return amount should have a paper trail.

Go from source rows to a generated return file. See what is included, what is excluded and what stops the package.

Click a return amount. Find its source rows.

A three-row synthetic sales ledger becomes a generated Virginia ST-1 / ST-1A text file. The production package writer and validator ran against the disclosed sample snapshot. This demonstrates package generation; ingestion, review approval and state acceptance are separate steps.

Gross sales: $150.00

VA-001 · General retail direct sale

Included: direct sale; general_retail identity and locality stipulated.

Source sale
$100.00
Seller tax
$5.30
VA-002 · Documented resale

Included as exempt: synthetic reviewed certificate stipulated.

Source sale
$50.00
Seller tax
$0.00

The downloadable ledger, package snapshot and file preserve these amounts. No customer data is used.

Where did the marketplace sale go?

VA-003 retains $200 of source sales and $10.60 of stipulated facilitator-remitted tax in the audit trail. It contributes $0 to this seller’s return. The ledger’s $350 of source sales reconciles to $150 on the seller return plus $200 facilitated sales.

One unresolved amount stops the package.

In a second execution, the snapshot includes $2.65 of unresolved marketplace responsibility. The same writer refuses to produce a file.

unresolved marketplace responsibility blocks ST-1

The sample account is deliberately synthetic. The generated file has never been filed, transmitted, accepted by Virginia or paid. Bank fields are empty and payment is zero.

Review the supported scope and exclusions
  • cigarette sales (field 6 zero-filled)
  • food & personal hygiene reduced-rate sales (fields 11-12 / ST-1A 5-7 zero-filled)
  • fuel for domestic consumption (fields 17-18 / ST-1A 9 zero-filled)
  • prepaid wireless E-911 fee (fields 19-20 zero-filled)
  • disposable plastic bag tax (fields 21-22 / ST-1A 8 zero-filled)
  • consumer use tax (fields 9/13 zero-filled)
  • dealer's discount (field 24 zero-filled — never auto-claimed)
  • penalty/interest (fields 25-26 zero-filled)
  • payment/banking (fields 28-30 never carry bank data)

Only the shown general-retail, locality-resolved seller snapshot is demonstrated. Required real registration, period review, channel evidence and filing authorization remain part of a private evaluation.

Give this evaluation to your AI agent.

Let your agent inspect the evidence, choose five hard cases and show where the Sales Tax Engine or Sales Tax Compliance fits your work. Start with public synthetic examples—no account or sales call required.

  1. Hand over the brief. Copy it into your agent, or attach the downloaded file.
  2. Choose your module. Check calculations and API behavior, or trace a compliance package to its source rows.
  3. Connect when you’re ready. A compatible MCP client can use the module tools you authorize. Your account’s access, pricing and approval controls apply.
Download for my agent

Read the brief

The brief gives instructions, not account access. Public examples are bounded demonstrations; connected tests need authorized access and may incur usage charges. See how agent connections work · Compare task rates and module plans.

Review it independently.

Use the same disclosed inputs and record failures as carefully as successful results. The public reproduction needs no account. A private API evaluation requires authorized sandbox access; the local reference environment is not a claim about every live configuration.

Publish your own findings with the pack version, date, limitations and any relationship to Prophit.ai. An untested competitor result stays untested. Continue to the API quickstart or check current pricing.

Run the public examples again.

No account or key needed. This runs the address lookup and package writer afresh with the disclosed inputs. It does not run your own transactions or submit a return.

Public reproduction script · Instructions

Evidence you can inspect. Limits you can see.

The business data is synthetic. The address example uses a small official-derived historical Illinois sample. The other executions use an isolated sandbox database, disclosed inputs and the same service implementations used by the application. No customer records, filings, payments or competitor credentials were used.

The pack distinguishes local component execution from hosted API execution. The optional replay script records your own sandbox results, including failures; it does not recreate the private reference database or certify a provider. Confirm supported coverage and effective dates before evaluating a real workflow.

Sales Tax Engine · Sales Tax Compliance · API coverage contract · Current pricing