# ISS-ZIWEI-2026-001 — Historical Ziwei defects and patches from August 2026

> **Identifier notice:** 001 now documents the historical audit/patch cycle. The natal palace/ruler correction published on 2026-11-17 has moved to [ISS-ZIWEI-2026-002](/issues/ISS-ZIWEI-2026-002/en/). This dossier does not restore the withdrawn audit or its Approved status.

## Abstract

This dossier records five historical defect groups, separates source claims from reproducible Git evidence, and documents incorrectly introduced ruler fields and Body residence. A duplicate-function cleanup leaves relative star placement unchanged at twelve positions. The dossier remains open pending bounded revalidation of each historical claim.

## 1 Scope

Opened 2026-11-18 for events on 2026-08-11 and 2026-08-12; the opening date is not backdated. Package: `QP-ZIWEI-001@2.1.0`. A historical claim of a successful patch is not an acceptance result.

## 2 Sources and methodology

The primary implementation witnesses are [ffa76720](https://github.com/the-quizzman/Quizzman-Lab/commit/ffa76720a2a33e5a0cc4226235bac7adfbfb93d6) and [348794c6](https://github.com/the-quizzman/Quizzman-Lab/commit/348794c6e9e1ee503f3af860e1f9308cfb35051a). [b446136](https://github.com/the-quizzman/Quizzman-Lab/commit/b446136124c9aad1971c89d4e86a486a891c3161) proves publication of historical claims, not their correctness. The reproduction reads exact Git revisions into a VM without checkout, production mutation or astronomical conversion. [Evidence JSON](/papers/QP-ZIWEI-001/artifacts/ISS-ZIWEI-2026-001-evidence.json) and [reproduction script](/papers/QP-ZIWEI-001/artifacts/verify-historical-provenance.cjs) are supplied.

## 3 Historical findings

### 3.1 UNVERIFIED — Fourteen principal stars

The old report claimed incorrect initial Ziwei and principal-star placement. In the inspected 11 August commit, anChumTuVi loses an earlier duplicate definition. Actual relative-star outputs before/after are identical at 12/12 positions because the later definition already overrides the earlier one. This cleanup alone does not prove a behavioral star-placement fix. The complete initial-placement history and all day/bureau combinations are not revalidated here. Severity: HIGH.

### 3.2 UNVERIFIED — Flanking topology

The historical checkPatterns claim concerns false positives when both helper stars occupy one adjacent palace. Distinct adjacent palaces must be checked independently. The entire historical implementation/fix pipeline has not been rerun for acceptance in this dossier. Severity: HIGH.

### 3.3 UNVERIFIED — Monthly transits

The report claimed a fixed Yin first-month palace was replaced with Dau Quan progression from annual Tai Sui, backward by birth month and forward by birth hour. The full historical buildLuuNienForYear pipeline, leap months and year/month boundaries remain unverified here. Severity: HIGH.

### 3.4 UNVERIFIED / RULESET_DIFFERENCE — Canh and Nham transformations

The historical diff changes Canh in core to Sun–Wu Qu–Moon–Tian Tong and Nham to Tian Liang–Ziwei–Tian Fu–Wu Qu. The southern table retains Zuo Fu for Nham Hua Ke. These are confirmed edits, not universal school-independent acceptance. School differences must be explicit. The existing THIEN_LUONG/Canh fixture-code discrepancy remains unverified in dossier 002.

### 3.5 BUG_CONFIRMED — Newly introduced ruler and Body-residence fields

Synthetic input: lunar 25/5/2006, 13:00, female. Before 348794c6, Life is Hai and Body is Chou; ruler and functional Body-residence fields are absent. After it, they are Wen Qu, Tan Lang and PHU_THE. Dossier 002 verifies Ju Men, Wen Chang and PHUC_DUC for the same input. The new ruler tables introduced wrong data, and the new Body label consumed a palace map already reversed before the commit. Severity: HIGH. The implemented correction belongs to ISS-ZIWEI-2026-002; this publication does not modify the engine.

## 4 Results and closure criteria

The provenance comparison passes, and twelve relative-star positions remain unchanged across the duplicate cleanup. This is bounded evidence, not acceptance of all five algorithm groups. The dossier SHALL remain open until each group has explicit rules, implementation revision, independent vectors and rerun results. Closure of 002 SHALL NOT imply closure of 001.

## 5 Limitations

Historical claims and computed results are separate. The removed audit is not republished as a correctness certificate. Synthetic birth data is not the original screenshot record. Full astronomical conversion, leap-month semantics, Zi rollover, transit layers and predictive efficacy are outside the verified scope.

## 6 Conclusion

The historical issue is formalized without reusing unsupported acceptance claims. Confirmed natal-foundation repairs and production evidence remain traceable in dossier 002.

## Bibliography

- [REF-ZIWEI-GIT-FFA7672] https://github.com/the-quizzman/Quizzman-Lab/commit/ffa76720a2a33e5a0cc4226235bac7adfbfb93d6
- [REF-ZIWEI-GIT-348794C] https://github.com/the-quizzman/Quizzman-Lab/commit/348794c6e9e1ee503f3af860e1f9308cfb35051a
- [REF-ZIWEI-GIT-B446136] https://github.com/the-quizzman/Quizzman-Lab/commit/b446136124c9aad1971c89d4e86a486a891c3161 (historical source statement only)
