跳转至

Historical version

This page records design, operation, or validation facts from an earlier release. It is not current V7.17.3 guidance. Start from the documentation hub for current instructions.

V7.15.1 Validation Report

The complete local validation on 2026-10-09 passed: 1,147 package tests PASS, one POSIX-specific check skipped on Windows, and 231 runtime tests PASS. Every capability mapping passed with no failures or errors. The report was collected at 2026-10-09T14:27:38Z.

Both language candidates are reproducible, with all 322 Plugin payload files in each archive verified individually. The account-level 7.15.1 Desktop Plugin was actually installed and passed installer verification; its cache matches the candidate payload. The current main model, effort and context settings were preserved, without restarting the application.

One real Desktop Explorer acceptance call successfully read the two specified files and confirmed the candidate version and nine-profile policy. The installed script approved, and the tool requested, gpt-6-sol / medium; scheduling metadata is not used to prove the underlying model identity. Native creation, file reads, terminal accounting and minimal metric receipts were read back.

The call settled as UNKNOWN according to actual lifecycle evidence. Including the earlier retained UNKNOWN call, the new-policy ledger totals eight planning units, two attempts and zero in flight. No refund or rewrite from UNKNOWN to PASS occurred. Actual credits remain unknown; historical research accounting and the earlier call remain intact.

The deterministic matrix covers nine profiles and three roles; real acceptance covers only the Explorer chain above, without claiming measured performance or statistical qualification for every model. This report establishes local code, artifacts and the stated Desktop scope. Commit, GitHub CI, tag, provenance and public-release facts are established separately by their remote records.