The evidence contract is the product.
One bounded workflow moves from named acceptance checks to reproducible evidence, one retest register and an explicit human-owned release state.
- Release scopeEnvironment, allowed-data rule, release owner and acceptance checks.
- Test matrixRisk, preconditions, steps, expected and actual behavior, state and evidence references.
- Issue registerStable issue IDs, reproduction, expected/actual behavior, severity and evidence.
- Retest registerOne bounded retest round linked back to the original issue.
- Release receiptReconciled counts and open issues resolved into PASS, REVIEW or BLOCKED.
- Offline validatorPython standard library only; no network calls, accounts or paid dependencies.
What the public evaluation proves
| Check | Evidence | State |
|---|---|---|
| Workflow boundary | One fictional scheduling workflow and explicit non-production boundary. | PASS |
| Cross references | Every test, issue and retest identifier resolves without duplication. | PASS |
| Visible exception | One reproducible issue remains open after a bounded retest. | REVIEW |
| Receipt reconciliation | Six cases, five PASS, one REVIEW and one open issue agree. | PASS |
| Commercial activation | No price, license, checkout, payment, tax or support term is active. | REVIEW |
Boundaries that stay visible
- Owned fictional data only; this is not a customer case or production result.
- No credentials, personal records, payment data, regulated data or live-account access.
- No security, performance, accessibility, legal, financial, medical or compliance certification.
- No automatic release decision; the accountable human owner retains the decision.
- Private-sector, non-government and non-official projects only.
The full template ZIP remains local and ready_not_active. The public file is an owned synthetic evaluation encoded as Base64 because the publishing connector accepts text only. It contains no blank commercial templates, active terms, checkout or payment collection.