BRING FIVE HARD TRANSACTIONS — INDEPENDENT REVIEW GUIDE
Published November 5, 2026. Reference evidence pack v1.0.0, executed October 21, 2026.
https://prophit.ai/bring-five-hard-transactions

The business records are synthetic. No client invoices, names, credentials or filings are included. The historical address sample is official-derived. A reference component execution is not a hosted end-to-end or state certification result.

1. Inspect the evidence without giving us your email.
Download https://prophit.ai/static/data/five-hard-transactions/v1/pack.zip
Read README.txt, manifest.json and each case's limitations before scoring.
Record the pack version, transaction date, execution environment and your review date.

2. Reproduce the public examples.
Download https://prophit.ai/static/data/five-hard-transactions/v1/reproduce.py
Review the script, then follow its --help and README instructions. It uses the public reproduction endpoint; it does not require an API key or use your own data. Preserve success, failure and source-hash mismatches. The address sample and return writer do not exercise the entire calculation workflow.

3. Evaluate the calculation engine.
https://prophit.ai/products/sales-tax-engine
https://prophit.ai/api-access/quickstart
The optional v1/replay.py uses your separately authorized sandbox access. Follow the pack's setup instructions; it does not create the private reference database for you. Never paste keys into an AI conversation, URL, review or public issue. Store credentials using your approved secret mechanism. Do not send customer transactions to the public reproduction endpoint.
Inspect missing product identity, certificate dates, credits, retry behavior, jurisdiction evidence and configuration differences. Test only the coverage actually agreed for your evaluation.

4. Evaluate compliance independently from calculation.
https://prophit.ai/bring-five-hard-transactions/compliance-package
Trace the four return fields and the marketplace exclusion to source rows. Inspect the unresolved-responsibility refusal. The generated Virginia ST-1 sample is marked SYNTHETIC-DO-NOT-FILE. It has never been filed, accepted or paid. Confirm real state/form/channel support, review workload and responsibility separately.

5. Record your own findings.
Use review-scorecard.csv. Keep untested results blank. Distinguish observed execution, documentary evidence, a claim needing verification and unsupported scope. Do not convert a source citation into proof of a product feature. Include defects, limitations and any vendor relationship in a published review. A competitor comparison requires authorized access and equivalent inputs; no competitor execution is supplied in this pack.

6. Price the actual workload.
https://prophit.ai/pricing is the current pricing source. Compare task rates and subscriptions separately. Include the number of locations, calculation volume, preparation and human review responsibilities. A private evaluation request does not enroll or charge you.

Request a scoped evaluation:
https://prophit.ai/bring-five-hard-transactions/switching-evaluation
Share workflow descriptions only. Arrange a private handoff for actual business records.