VSO announcements

VSO · · RESEARCH & DEVELOPMENT

VSO reports 23 verification, SDK and Explorer improvements

Twenty-three pull requests merged on 15 November are grouped into five areas. The separate header change is excluded. This reports research, implementation and documentation changes, not 23 new features or releases.

1. Explorer requests, navigation and API inputs

GUI #29 distinguishes pending, error, empty and offline-paused states, repairs same-page searches and filter offsets, and navigates using the API’s current-page count. API #31 rejects unsupported query parameters and preserves parser errors with 400/413/415 status.

The public GUI is updated from merged revision fb6a313. Ordinary dashboard routes remain synthetic. The separate SDK Evidence Demo retains its existing API connection; this GUI publication does not deploy API #31 or establish that the serving API uses that revision. Even intact synthetic evidence remains INDETERMINATE overall without verified external anchoring.

vcp-explorer-gui #29 · vcp-explorer-api #31

2. SDK and CAP-SRP evidence/output safety

SDK #18 rejects collisions between seal outputs, input evidence and key files before processing, including relative paths, symlinks and hard links. CAP-SRP #5 protects input evidence from report output, separates JSON from diagnostics, and propagates launch/output failures.

These preflight checks are not concurrency locks or atomic multi-file transactions. Evaluate the SDK repair at main revision 0832f1d, separately from the earlier installation snapshot 8eadded8, historical v0.1.0 tag and unchanged 0.1.0 package metadata. No new package release is announced.

cap-srp #5 · vcp-sdk #18

3. Configured public keys in research verification

VAP #72/#73, VCP #32, Evidect #43, Supervision #6 and TradingView #7 reject weak points and selected noncanonical encodings in caller-configured Ed25519 trust. Admitting such a weak key is a prerequisite; submitted evidence cannot thereby replace an existing strong trusted key.

The repairs do not authenticate trust configuration or claim complete prime-subgroup validation. Consult each PR for its bounded checks. Historical consumers and distributed source copies retain their pinned revisions; successful historical rechecks do not silently test current-code repairs.

vap-spec #73 · vap-spec #72 · evidect-site #43 · vcp-supervision-node-poc #6 · vcp-tradingview-rta-reference #7 · vcp-spec #32

4. Reference implementations and unresolved results

FIX, IBKR, Nasdaq, cTrader and Cloudflare references improve selected-evidence handling, malformed types, ambiguous JSON, empty records and CLI behavior. Payout, GDPR and Supervision PoCs distinguish incomplete comparisons, empty chains and unavailable checks from successful local results.

Local agreement does not prove authentic external observation, complete capture, payment completion, deletion or legal compliance. Each implementation retains its own verdict conditions and test scope.

vcp-payout-dispute-poc #6 · vcp-gdpr-poc #5 · vcp-nasdaq-rta-reference #6 · vcp-ibkr-rta-reference #6 · vcp-ctrader-rta-reference #3 · vcp-fix-rta-reference #5 · vcp-cloudflare-rta-reference #5 · vcp-supervision-node-poc #7

5. Evidence rechecks and saved-review research

Evidect #37 adds credential-lifecycle evidence, receiver/local-witness checks and saved reviews. Separate keys, databases and a child process remain under one experiment controller, not independent administration. Limited VALID results do not promote unresolved overall assurance into success.

Browser Recorder #178 adds a separate Python implementation for recomputing browser-produced signed evidence and rejecting invalid origin representations. Separate implementation is not independent third-party reproduction. It adds no external anchor, trusted time or completeness guarantee.

vcp-site #178 · evidect-site #37

Related documentation and CI maintenance

VAP #74 updates Actions runtimes in 11 workflows while retaining pinned dependencies and check conditions. .github #14 and vcp-docs #2 align setup requirements and SDK/Explorer entry points. These changes are included in the 23 PRs.

First-party CI and site publication checks are separate

At 20:57 JST on 15 November, all 55 workflow runs executed at the latest main revisions of 17 of the 19 reviewed repositories had succeeded. The two documentation-only repositories, .github and vcp-docs, had no corresponding CI. These are workflow runs, not test-method counts. This is a review of linked GitHub records, not independent reproduction. The CI, display and hash checks for this site publication are separate from the research checks.

GitHub Actions · 55 workflow runs

Current main, fixed distributions and history

Fixed tags, signed data, research ZIPs, K3 and current normative requirements are unchanged. The interop v0.2.0 known issue and main-only repair, and earlier 159-test/110-image-comparison records, retain their original revisions and conditions. The VAP specification SHA-256 remains a35d5e08…a19998; this publication neither revises nor adopts the specification.

Fixed interop release / known issue · Fixed research kit

The 23 merged changes and their revisions

ExplorerVCP SDK