Verification evidence
Use this index to choose a check and interpret its result. The 1.0.1 release uses escrow-v5 / head-v2. See its release notes and the external qualification record. Published RC2 and its release records use escrow-v4 / head-v1. The 2026-10-07 development verification separately records fresh escrow-v5/head-v2 funded runs. Neither is mainnet evidence.
| Evidence | What it establishes | Limits / reproduction |
|---|---|---|
| 1.0.1 release | 55 external scenario groups and a fresh funded 18-flow exact/batch plus five-stage hash-chain run using the exact installed package archives. | One Testnet RPC; controlled stress acceptance facts; archive hashes must match to preserve proof after documentation/frontend edits. |
npm run verify from protocol/ |
Local schemas, signed vectors, SDK behavior, persistence contracts, adapters, fixtures and public package contents. | Simulated network results; no funded chain acceptance. See entry point. |
| Pinned Rust consensus harness | Transaction and covenant acceptance/rejection under the recorded consensus source, including six signature scopes and independent guards. | Offline validation. Pins and commands: dependencies, recorded scope. |
check:covenant-fixtures |
Authenticated pinned SilverScript objects, locked isolated build, and independent escrow-v5/head-v2 compilation agree with fixtures/ABI and generated runtime modules, including constructor/state variants. JSON records compiler/tree, manifest/lock/toolchain/binary and compilation hashes. | Installed host tools are trusted; no funded transaction or mainnet acceptance. Local verify performs internal consistency only. Release reproduction. |
| x402 compatibility experiment | HTTP envelopes and trusted request binding against pinned @x402/core. |
Simulated settlement and experimental attestation; not an upstream SDK integration release. |
| RC2 release record and Testnet report | Historical funded exact, batch and recovery observations with recorded transaction IDs and source scope. | Preserve the original template versions and prerequisites. Reproduction guide. |
| 2026-10-07 development proof | Fresh escrow-v5 18-flow exact/batch proof and head-v2 paid HTTP/owner recovery, with transaction IDs and runtime fingerprints. | Reference ALL signing, one configured Testnet-10 PNN, local harnesses; no hosted deployment or upstream SDK settlement. |
| Hash-chain live record | Historical head-v1 evidence and reproduction commands for the current proof. | Keep template versions distinct; local issuer/reorg tests are separate evidence. |
check:address-transaction-compatibility |
Synthetic mainnet-address serialization and signing with an explicitly supplied Kaspa WASM module. | No real UTXOs, broadcast, network acceptance or mainnet readiness. |
| Host checks from repository root | Website package consumption, browser behavior and gateway persistence/Worker integration. | A Worker dry run or simulated browser payment is not deployed or funded proof. |
The threat model states trust assumptions; the store contract and lock contract state host obligations. Mainnet readiness owns the open deployment gates. Hash-chain research is a historical appendix, not the current integration contract. The upstream readiness record identifies contribution requirements and separates unresolved spec decisions from later SDK work.
Distinct failure suites remain separate: schema rejection, cryptographic and
transaction validation, exact replay/recovery, batch accounting/recovery,
client disclosure/funding limits, facilitator transport binding, and gateway
storage failure/rollback. Shared fixture setup is under test-support/; it
must not replace those assertions or turn mock signatures into security proof.
Source: /docs/evidence.md