# Public-Site Audit Findings, 2026-07-19 **Portfolio:** V1 **Reviewed public snapshot:** `e57e09128ca87b70f99f44e4387224ebda2ff14b+working-tree` **Source identity limitation:** this legacy dirty-tree label predates the working-tree fingerprint control. The exact dirty source tree was not archived and cannot be reconstructed from the label. The published artifact snapshot hash below remains the durable identity for what was reviewed. **Published snapshot hash:** `6f8003b32cb9ac7e7e3a39d7e98eca63197d039632e76abcd14c14beae34c00d` **Reviewer attribution:** not supplied for public use **Provenance:** supplied to the release owner; human or AI provenance was not established for the public record. **Intake:** a detailed externally supplied public-site and corpus audit reported reading the English corpus, inspecting three-language metadata, and testing the public research tools. This record splits its actionable findings so each can be dispositioned independently. The accepted corrections were implemented in source commit `9012e002c654de5ff5ac45c3ed4bdd43645cc876` and website commit `2d862ea33af971c4e6b9ee7bdc7b9ba4a37dd41c`. The [internal verification record](../evidence/v1-preview-internal-verification-2026-07-20.md) records the clean multilingual build, artifact checks, application tests, dev and production deployments, and live verification. That evidence closes `FB-V1-2026-001` and `FB-V1-2026-003` through `FB-V1-2026-010` as implementation findings. It does not claim specialist endorsement of the underlying conclusions. `FB-V1-2026-002` remains open and continues to block stable promotion. ## FB-V1-2026-001 **Priority and category:** P1, doctrinal propagation **Finding:** current DDF ranks staged conditional final destruction as its preferred terminal synthesis and universal restoration as a lower-confidence permitted hope, while *Rethinking Reality* still called restoration DDF's preferred completion. **Disposition:** accepted. The governing hierarchy is propagated directly through English, Spanish, and Indonesian *Rethinking Reality*. The affected backlog essays and historical alignment records are corrected or explicitly superseded, and a regression check protects the hierarchy. **Verification:** full multilingual builds, exact generated-text checks, public book regeneration, and deployment verification are required before closure. **Closure:** closed for doctrinal propagation consistency after the clean three-language build, public regeneration, deployment, and live verification. Formal theological review remains open under `FB-V1-2026-002`. ## FB-V1-2026-002 **Priority and category:** P1, review and attribution **Finding:** chapter records used `not_specified`, the interface paraphrased that state ambiguously, and the wider corpus had not completed formal external specialist review. **Disposition:** accepted with a hard boundary. The manifest now records the actual author, editorial owner, Gabriela Tifany's Indonesian editorial and cultural-review role specifically for the Indonesian edition of *Rethinking Reality*, `not_reviewed` status for the wider corpus, required specialist domains, and the fact that external review blocks stable promotion. The newer companion localizations remain pending native-speaker review. A public review matrix defines completion evidence. The site and machine-readable corpus expose the same state. **Remaining dependency:** qualified external people must perform the work. Software, AI agents, and the release owner cannot truthfully mark that human gate complete. **Status:** open. The metadata, public disclosure, and stable-release control are implemented and verified, but the required external reviews have not been completed. ## FB-V1-2026-003 **Priority and category:** P2, source apparatus **Finding:** practical books promised more recoverable claim-level research than the public site exposed. **Disposition:** accepted. The five existing source and conclusion ledgers are current public source apparatus rather than future plans. They are copied into versioned downloads, linked from per-book research pages, incorporated into source and AI bundles, and associated with exact hashes and a correction path. **Closure:** closed after publication validation and live verification of the per-book research routes and four versioned packages. ## FB-V1-2026-004 **Priority and category:** P2, research method and interface **Finding:** CRM is a proposed integrative protocol rather than a validated diagnostic or predictive model, and the live grid described more variables than it exposed. **Disposition:** accepted. Public copy consistently labels CRM as an early proposed protocol. The grid exposes prediction error, meaning gap, source trust, capacity, and agency; shows the resulting profile; stops interpretive use at the red-capacity boundary; and avoids diagnosis or validated-score claims. **Closure:** closed after application tests and deployed browser verification of the five variable groups, pilot language, live result, and keyboard path. ## FB-V1-2026-005 **Priority and category:** P2, engineering **Finding:** full-book references such as `Romans 5:12` could fail while abbreviations worked, and generic text relevance could outrank the exact verse. **Disposition:** accepted. One shared 66-book parser now handles English, Spanish, Indonesian, abbreviations, OSIS forms, punctuation, and numbered-book ambiguity. Exact references bypass generic search readiness, and exact matches rank before generic pagination. **Closure:** closed after the frontend unit suite, production build, and live Passage API verification passed. ## FB-V1-2026-006 **Priority and category:** P2, data quality **Finding:** relational person records entered the Dictionary as lexical headwords, and extended Strong identifiers could be misrepresented as ordinary forms. **Disposition:** accepted. Builder, API, detail, and UI boundaries reject the known relational pseudo-headwords. Extended identifiers have a separate typed field and localized label. Builder validation rejects recurrence. The production data store is not rewritten as part of this correction; safe read-time filtering makes the correction effective without violating the completed-migration data freeze. **Closure:** closed after builder and frontend tests, publication checks, and live dictionary verification passed without a production table rewrite. ## FB-V1-2026-007 **Priority and category:** P2, source and production **Finding:** the Shannon PDF URL in DDF was corrupted during LaTeX-to-Markdown processing because the literal tilde was lost or separated. **Disposition:** accepted. URL arguments now have a dedicated conversion path that preserves URL characters and removes formatting whitespace. Publication validation asserts the exact URL in all three generated DDF editions. **Closure:** closed after all three DDF editions built and publication validation passed against the generated public data. ## FB-V1-2026-008 **Priority and category:** P2, philosophical **Finding:** the worldview comparison could count Christian commitments as shared data, use ambiguous `DDF surplus` language, compress naturalistic moral realism, and slide among moral truth, reasons, motivation, enforcement, and final judgment. **Disposition:** accepted. DDF now distinguishes shared public explananda from confessional commitments, separates the moral questions, presents a serious naturalistic-realist alternative, and labels its additional Christian commitments and integration without treating disagreement as unexplained data. **Closure:** closed as an implementation finding after the clean DDF build and publication verification. External philosophical review remains open under `FB-V1-2026-002`. ## FB-V1-2026-009 **Priority and category:** P2, exegetical **Finding:** disputed biblical areas needed fuller engagement with the strongest opposing readings, especially sexuality and Adam. **Disposition:** accepted. DDF now represents the ancient-category, exploitation, status, pederasty, idolatry, Romans 1 to 2, lexical-limit, and canonical-trajectory arguments before stating why it retains its traditional synthesis. Its Adam section now distinguishes recent sole-genetic-progenitor, genealogical, archetypal or corporate, and historical covenant-head accounts, and marks the line between genetic findings and theological inference. No conclusion is strengthened merely by listing lexical or scientific facts. **Closure:** closed as an implementation finding after the clean DDF build and publication verification. External exegetical and scientific review remains open under `FB-V1-2026-002`. ## FB-V1-2026-010 **Priority and category:** P2, accessibility and interface **Finding:** beta research tools needed truthful maturity labels, matching controls, and dependable loading, empty, error, keyboard, focus, and live status behavior. **Disposition:** accepted. CRM is labeled a pilot tool and uses native labeled controls, grouped agency choices, an accessible plot description, and a live result region. Passage and Dictionary expose discovery-aid boundaries and explicit data states. Focused automated tests cover exact-reference and dictionary-boundary behavior; release validation and deployed keyboard checks remain required before closure. **Closure:** closed after release validation, the 109-test frontend suite, and a deployed keyboard pass reached the skip link, labeled CRM controls, agency choices, scenario buttons, and polite live-result region. This internal check is not a claim of independent WCAG certification.