v3.8.2

[3.8.2] - 2026-07-07

Patch [clarification] — GitHub Copilot changelog refresh: Model Compatibility snapshot brought current (5 new models, 3 deprecation promotions, 1 currently-Primary downgrade); welcome-baseline $comment extended with enterprise-managed-settings.json per-OS paths.

Docs-only sibling to v3.8.1 (which shipped the VS Code 1.126–1.128 platform-awareness refresh). No rule changes, no settings object changes, no manifest contract change.

Source: Tier D items D2 + D3 from Alex_ACT_Supervisor/docs/proposals/tier-d-implementation-menu-2026-07-07.md, tracing back to copilot-changelog-2026-07-07.md Tier A.

Changed

• README.md § Model Compatibility § Snapshot — validation date 2026-05-26 → 2026-07-07. Deprecation flag updates: GPT-4.1 ⚠️ (retiring 2026-06-01) → ❌ (deprecated 2026-06-02); GPT-5.2 + GPT-5.2-Codex ⚠️ (retiring 2026-06-01) → ❌ (deprecated 2026-06-05); Claude Opus 4.6 (fast mode) recommendation downgrade ✅ Primary (inferred) → ⚠️ (deprecation announced 2026-06-18); Gemini 2.5 Pro ✅ Primary → ⚠️ (deprecation announced 2026-07-02); Gemini 3 Flash 🟡 Utility → ⚠️ (deprecation announced 2026-07-02). Recommendation legend ❌ Do not adopt reworded from Retiring 2026-06-01 per GitHub Docs → Deprecated per GitHub Copilot changelog.
• .github/config/welcome-baseline.json ($comment only, settings block untouched) — extended Category (8) block with an "Enterprise-managed environments" note documenting VS Code 1.125+ MDM channel + VS Code 1.127+ file-based delivery (GA 2026-07-01) + per-OS paths for managed-settings.json (Windows / macOS / Linux). Precedence order clarified: MDM + file-based delivery both override user-scope settings.json.

Added

• README.md § Model Compatibility § Snapshot — 5 new model rows: Claude Sonnet 5 (✅ Primary inferred, GA 2026-06-30), Claude Fable 5 (✅ Primary inferred, GA 2026-06-09), Claude Opus 4.8 fast-mode Preview (⚠️ Test first, 2026-06-29), Kimi K2.7 Code (⚠️ Test first, GA 2026-07-01), MAI-Code-1-Flash (⚠️ Test first, GA 2026-06-26 Business/Enterprise).
• README.md § Model Compatibility — new paragraph below the Snapshot table: "Note on Auto mode + configurable reasoning" documenting Auto mode GA 2026-06-17 (legitimate alternative to explicit Primary-slot pattern; not yet validated on ACT discipline; recommend explicit selection until measured) and configurable reasoning levels + larger context windows (GH 2026-06-04, extends same-name model rows).

Brain contract

Unchanged: min_extension_version stays 9.5.1, brain_subtrees stays [.github], marker_schema stays v2, manifest spec stays 1.4. Only edition_version and generated_at change in the manifest.

Heir upgrade

Standard /upgrade (Extension ACT: Upgrade Brain). No --allow-major needed. Heirs picking models via UI see the Snapshot table's new/deprecated rows on next visit to README.md; behavior gate for existing rules unchanged.

Falsifier

2026-08-07 (30 days per Copilot proposal falsifier):

• (a) A recommendation flag for a new model (Sonnet 5 as ✅ Primary, Fable 5 as ✅ Primary, Kimi K2.7 / Opus 4.8 fast / MAI-Code-1-Flash as ⚠️ Test first) proves miscalibrated when a heir reports the model doesn't fit architectural-needs criteria → flip the flag via v3.8.3 patch.
• (b) Opus 4.6 (fast) deprecation gets rescinded before it takes effect → restore the ✅ Primary flag.
• (c) A model marked deprecated shows up in the Copilot picker still available past its deprecation date → clarify the flag semantics or amend the row.
• (d) The Auto mode + configurable reasoning paragraph produces heir confusion about "should I pick a specific model or use Auto?" → expand the paragraph or drop it.