VSO / Research resources

VSO · Public research resources

Payment API authorization and evidence research

Mock experiments distinguish valid signatures, exact-request approval, known execution outcomes and retained evidence.

Public-source review: · Informative, non-normative

Four retrospective evidence questions

The offline verifier evaluates integrity, authorization, execution match and control state separately. VALID does not mean a payment settled: evidence can correctly match a failed or reversed outcome.

INDETERMINATE retains missing evidence, trust, freshness or coverage. This verifier does not authorize future payments. See Offline payment evidence experiment.

Separate execution, authorization and collection

  1. Runtime: loopback HTTP and SQLite toy balances exercise durable claims, one dispatch and read-only reconciliation. The base runtime does not verify business approval. Controlled payment runtime pilot
  2. Authorization: a separate pilot binds synthetic maker/checker approval to the exact request. It is a controlled application route, not an exclusive boundary around the provider. Controlled payment authorization pilot
  3. Collection: another process stores original admission records and provider receipts. Evidence recovery does not resend the payment. Controlled payment collector pilot

Unknown outcome and missing custody are different

If a provider acts but its response is lost, retain UNKNOWN rather than assuming failure. A later read-only reconciliation can establish a mock outcome; receipt custody remains a separate question.

No bank, real funds or human approval system is connected. Separate processes remain under the same operator and host user; they do not establish independent custody or OS isolation.

Recheck a historical run

Preserved 6 November collector run preserves the original ZIP, public evidence, source and records of one first-party CI run on 6 November. Rechecking those saved bytes is neither a fresh run nor independent third-party reproduction.

Sources and reviewed revisions

This page summarizes public materials at pinned commits. Creating it is not a new research test run. Historical test records and fixed distributions remain unchanged.