[mirror29]

This records three explicit prospective legacy fact reviews and a fresh owner coverage check, without treating either as full E2 acceptance.

• Individually previewed 31 active legacy regulatory facts. The initial write corrected only the explicitly inspected archived Kraken SoFiUSD educational article. A separate helper process explicitly replayed the frozen request and returned the same audit/replacement/retirement IDs with created=false.
• Subsequently inspected the two BTC legacy regulatory risks and retained both regulatory classifications, appending one v3 confirmation for each without retirement. The SEC source only archives a 76-character title; full article truth is not certified. Total audits are now 3. These confirmations occurred after the recorded owner matrix; post-confirmation owner re-preflight remains pending.
• Independently checked the frozen source hash and the single logical source's before/after SQL visibility. Original v2 versions, evidence and effective time remain intact; correction availability is the current database time.
• The logged-in owner's normal suggested-window matrix at 2026-10-14T03:14:24.850Z reports BTC/ETH/SOL upper bounds 3/8, 0/8 and 1/8. No asset was selected, approval submitted or E2 model called. The October 13 and October 14 selection windows differ; this is not a controlled coverage regression comparison. The quality guard blocks execution but does not subtract from the reported raw upper bound. BTC additionally reports two legacy regulatory provenance risks, not two confirmed semantic errors.
• English desktop preparation preserves the SOL target, hides the result after a real cutoff-hour change, and returns the same insufficient result when the suggested window is restored. Mobile rendering, server-side stale-approval rejection and execution remain unverified.

Validation

• Cross-file consistency: 19 passed, 0 failed, 12 existing warnings; git diff --check clean.
• Independent read-only incremental documentation review: no confirmed new P1/P2, limited to sanitized outputs and reported CUA observations.
• Deployed Data/Research remain 9fdb8b8; Dashboard/Mastra/Evolver remain 76727f6. No service restart or deployment in this acceptance run.

Documentation

• Current state now distinguishes the latest matrix from the October 13 historical sample and records the new production evidence.
• Data and Research READMEs describe the demonstrated helper-to-Data write and explicit-replay scope.
• Three new receipts preserve observations and limitations. Public preview metadata omits unneeded archived news titles/content. Public-diff redaction scan has no findings. Existing deployment receipts remain historical.
• No public API or architecture changes. This evidence does not establish authenticated incoming Research HTTP, production service restart/response-loss recovery, matched-control/FDR success, champion, Forward, holdout or adoption.

[cloudflare-workers-and-pages]

Deploying inalpha-web with    Cloudflare Pages

Latest commit:	82bc67a
Status:	✅  Deploy successful!
Preview URL:	https://c283370b.inalpha-web.pages.dev
Branch Preview URL:	https://codex-event-fact-single-prod.inalpha-web.pages.dev

View logs

[github-actions]

Reviewed PR head: 82bc67a73030

🤖 DeepSeek Code Review

Docs-only change: adds three 2026-10-14 validation receipts (owner free matrix/EN preparation, single Kraken correction with explicit replay, two BTC same-class confirmations) and updates docs/04-current-state.md plus Data/Research README status text accordingly.

Critical	Major	Medium
0	0	0

Findings

No medium-or-higher issues found in the supplied diff.

Review scope and limitations

Reviewed the supplied diff and project rules. No tests were executed by this reviewer.

• Diff is documentation/receipts only; no code, schema, config or API contract changed, so correctness/security/performance dimensions largely do not apply.
• Cannot validate the receipt JSON against a schema, nor confirm the production DB state (fact IDs, audit counts, v2/v3 version counts) beyond internal consistency of the diffed files.
• Could not verify referenced pre-existing receipts (e.g. event-fact-review-deployment-2026-10-14.json) still exist/unchanged since they are outside this diff.
• The e2-owner-matrix-language.json field original_strategy_window_display first element is 2025-10-09T06:00:00Z while the second is 2026-10-09T06:25:54.528Z (a 1-year span); plausible as a 1-year default strategy window, but the start year could not be confirmed as intentional from the diff alone.
• Whether appending a v3 same-class confirmation should clear the legacy v2 quality guard is asserted as 'not cleared' in prose only; underlying guard logic is not in this diff.

[mirror29]

Review response

Finding	Resolution	Evidence / limits
MEDIUM: coverage samples not reconciled	Addressed in f4fb586: record both selection-window sets, sample times and release versions; explicitly mark these as different frozen inputs rather than a controlled regression comparison.	October 13 public receipt and current CUA observations agree. Per-event contribution differences remain unreconciled. inspect_event_coverage computes the raw upper bound before appending quality blockers, without subtracting legacy risks.
MEDIUM: acceptance title overstates readiness	Title updated to docs(e2): record fact reviews and blocked owner preflight.	Necessary inputs remain insufficient. The observed matrix predates the two BTC confirmations; post-confirmation owner re-preflight, mobile, approval, five generations and long-term validation remain unverified.

Latest-head CI and concrete review will be checked before merging this docs-only PR. No service rebuild is needed.

[mirror29]

Incremental review response

Finding	Resolution	Evidence / limits
MEDIUM: zero freshness lag could be mistaken for wall-clock age	Renamed the zero value to lag_vs_required_bar_open_seconds and added explicit bar-open age 4464.850s and bar-close age 864.850s.	Sample 03:14:24.850Z; last closed bar opens 02:00 and closes 03:00. No zero-age claim.
MEDIUM: v3 35→38 count not reconciled	Added independent read-only production reconciliation: baseline 35 + 1 completed normal extraction + 1 retired version + 1 replacement version = 38.	One cross-class correction operation creates two versions; replay creates zero. Normal fact 697872c9 was created at 02:31:07.616934Z. Later BTC confirmations are in a separate chronological receipt.

Resolved in 8c278ee. Independent incremental doc review found no confirmed new P1/P2 within sanitized evidence scope. Consistency remains 19 passed / 0 failed / 12 existing warnings. Final exact-head CI and concrete review are pending; no services will be rebuilt for these receipts.

[mirror29]

Retirement-ID clarification

The immutable source row remains fact 8a0ef52e (version 1, retracted=false). The operation appends e8cf1c77 as version 2 under the same fact_key/raw ID/evidence/effective time, with retracted=true, and appends d3df6f9d as the other-class replacement under its canonical distinct key. Both source_fact_id and retired_fact_id are market_event_facts.fact_id UUIDs; retired_fact_id identifies the new retirement version rather than an in-place mutation of the source. Active before/after lists intentionally exclude the retracted version.

82bc67a adds this explicit mapping from independently read-only-verified production rows and explains it in the Data README. Source code appends retirement through write_fact before inserting the linked immutable audit. Independent incremental doc review found no confirmed new P1/P2 within this evidence scope. Latest-head CI/review will be checked before merge; no production rebuild.