1C ERP Diagnostics v0.3.8 — Deterministic Accounting and State Integrity

1C ERP Diagnostics v0.3.8 — Deterministic Accounting and State Integrity

Overview

Version 0.3.8 adds executable accounting, effect-classification and case-state controls derived from the reviewed 1C cost-allocation case. It preserves the evidence-first Gate 0–10 contract: arithmetic validation can reject inconsistent data, but it cannot establish a 1C root cause without independent Gate 7 evidence.

Accounting and state controls

• exact fact/plan equality, share sum, deterministic allocation, residual, row coverage and observed-allocation checks without float arithmetic;
• independent before/after flags for completeness, allocation proportion, analytic key, cardinality and no material change;
• explicit business_basis_required when a proposal changes allocation proportions;
• executable cross-entity identity, run/input/tool/output freshness, lineage and downstream invalidation validation;
• MXL/XLSX property trees remain metadata until explicitly bound row evidence is available.

Security and regression coverage

• bounded fail-closed XLSX/ZIP metadata and package scanning, including local/central header differences, credentials and user-machine paths;
• release-tree and complete reachable-history checks for unsafe paths, credentials, containers and symlinks;
• 10 new accounting/state/security regressions, expanding the validated synthetic EVAL_RESULT_JSON suite to 26 cases;
• the synchronized v0.3.8 candidate passes 143 repository unit and regression tests locally.

Preserved ecosystem

• 32 packaged skills and marketplace ID one-c-erp-diagnostics-marketplace;
• Unica aefc880f9bab606a5c55ed11af563b740054a549 (v0.12.0);
• 1C Skills PowerShell 8cb7868145281d8e353831512cc1ffa72f1b5c89;
• 1C Skills Python c1f79f5ac9f31c620b8508f75464f8c42c559ae4;
• approved Velis assets and independent-project trademark boundary.

Validation

• Python 3.10/3.12, CodeQL, package, marketplace, deterministic lock, publication-history and all 143 repository tests: PASS.
• Exact installed candidate clean-session acceptance: 26/26 and RUNTIME ACCEPTANCE: PASS.
• Public marketplace main at 826aae46ed278e42c182c1be3e3c93cc2a53766c: 26/26 clean sessions and RUNTIME ACCEPTANCE: PASS.
• Runtime evidence establishes conformance to the synthetic Gate 0–10 contract; it does not establish a real 1C root cause without case-specific evidence and independent Gate 7.