docs: mirror proper-lint-fixes rule into Copilot instructions

.github/copilot-instructions.md

@@ -28,7 +28,7 @@ A companion guide `CLAUDE.md` at the repo root carries the shared rules for Clau
 - Updating documentation
 - Writing or improving unit tests
 - Code refactoring within established patterns
-- Fixing linting/formatting issues
+- Fixing linting/formatting issues — **proper lint fixes only**: fix the cause. Never use disable comments (`eslint-disable`, `noqa`, `type: ignore`, or similar) unless explicitly approved by the repository owner.
 - Updating dependencies (after checking security advisories)
 
 **Tasks requiring human review:**

changelog.d/copilot-proper-lint-fixes.md

@@ -0,0 +1,7 @@
+### Changed
+
+- **Copilot gets the "proper lint fixes only" rule.** The lint task line in
+  `.github/copilot-instructions.md` now says to fix the cause and never add
+  disable comments (`eslint-disable`, `noqa`, and similar) without the owner's
+  explicit approval, matching `CLAUDE.md`; the instruction tests pin the rule on
+  both sides. (#12190)

tests/unit/test_agent_instructions.py

@@ -317,6 +317,13 @@ def test_every_named_skill_exists() -> None:
     ],
     # Owner decision 2026-10-15: no `pip` fallback for either audience.
     "python tooling is uv only": ["`uv` only", "never `pip`", "instead of falling back to `pip`"],
+    # Owner decision 2026-10-15. The pin carries the prohibition and its one
+    # exit (explicit approval), not just the topic "lint".
+    "proper lint fixes only, no disable comments": [
+        "proper lint fixes only",
+        "never use disable comments",
+        "unless explicitly approved",
+    ],
     # The repository's most consequential rule, and the one an agent that opens
     # and edits PRs is most able to break: specs and implementations go through
     # the workflows, and their PRs are merged by `impl-merge`, never by hand.