test: add dedupe conflicts non-utf8 contracts

CHANGELOG.md

@@ -6,6 +6,8 @@ This project has a published GitHub Release line, but no stable support or API g
 
 ## [Unreleased]
 
+- Add non-UTF-8 error-contract coverage for `dedupe` and `conflicts`.
+
 ### Added
 
 - Added a consolidated post-audit action plan for the v0.4.0 hardening and release-preparation sequence.

tests/test_cli.py

@@ -268,6 +268,32 @@ def test_conflicts_returns_two_for_symlinked_instruction_file(self) -> None:
             "conflicts: AGENTS.md\n",
         )
 
+
+    def test_conflicts_returns_two_for_non_utf8_instruction_file(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp_dir:
+            root = Path(tmp_dir)
+            original_bytes = b"# Agent instructions\n\n\xff\xfe\n"
+            agents_file = root / "AGENTS.md"
+            agents_file.write_bytes(original_bytes)
+
+            stdout = io.StringIO()
+            stderr = io.StringIO()
+
+            with redirect_stdout(stdout), redirect_stderr(stderr):
+                exit_code = main(["conflicts", str(root)])
+
+            self.assertEqual(agents_file.read_bytes(), original_bytes)
+
+        self.assertEqual(exit_code, 2)
+        self.assertEqual(stdout.getvalue(), "")
+        self.assertEqual(
+            stderr.getvalue(),
+            "ERROR: instruction file is not valid UTF-8 and cannot be checked "
+            "for conflicts: AGENTS.md\n",
+        )
+        self.assertNotIn("\\xff", stderr.getvalue())
+        self.assertNotIn("255", stderr.getvalue())
+
     def test_dedupe_reports_duplicate_lines(self) -> None:
         with tempfile.TemporaryDirectory() as tmp_dir:
             root = Path(tmp_dir)
@@ -297,6 +323,32 @@ def test_dedupe_reports_duplicate_lines(self) -> None:
         self.assertIn("AGENTS.md:3", text)
         self.assertIn("CLAUDE.md:3", text)
 
+
+    def test_dedupe_returns_two_for_non_utf8_instruction_file(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp_dir:
+            root = Path(tmp_dir)
+            original_bytes = b"# Agent instructions\n\n\xff\xfe\n"
+            agents_file = root / "AGENTS.md"
+            agents_file.write_bytes(original_bytes)
+
+            stdout = io.StringIO()
+            stderr = io.StringIO()
+
+            with redirect_stdout(stdout), redirect_stderr(stderr):
+                exit_code = main(["dedupe", str(root)])
+
+            self.assertEqual(agents_file.read_bytes(), original_bytes)
+
+        self.assertEqual(exit_code, 2)
+        self.assertEqual(stdout.getvalue(), "")
+        self.assertEqual(
+            stderr.getvalue(),
+            "ERROR: instruction file is not valid UTF-8 and cannot be "
+            "deduplicated: AGENTS.md\n",
+        )
+        self.assertNotIn("\\xff", stderr.getvalue())
+        self.assertNotIn("255", stderr.getvalue())
+
     def test_dedupe_returns_one_when_no_instruction_files_are_found(self) -> None:
         output = io.StringIO()
 

tests/test_conflicts.py

@@ -125,6 +125,19 @@ def test_reports_all_current_rule_family_conflicts(self) -> None:
             self.assertEqual([location.path for location in group.allow_locations], ["AGENTS.md"])
             self.assertEqual([location.path for location in group.block_locations], ["CLAUDE.md"])
 
+
+    def test_rejects_non_utf8_instruction_files(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp_dir:
+            root = Path(tmp_dir)
+            original_bytes = b"# Agent instructions\n\n\xff\xfe\n"
+            agents_file = root / "AGENTS.md"
+            agents_file.write_bytes(original_bytes)
+
+            with self.assertRaisesRegex(ValueError, "not valid UTF-8"):
+                build_conflict_report(root, discover_instruction_files(root))
+
+            self.assertEqual(agents_file.read_bytes(), original_bytes)
+
     def test_rejects_symlinked_instruction_files(self) -> None:
         with tempfile.TemporaryDirectory() as tmp_dir:
             root = Path(tmp_dir)

tests/test_dedupe.py

@@ -42,6 +42,19 @@ def test_ignores_short_boilerplate_lines(self) -> None:
 
         self.assertEqual(report.duplicate_group_count, 0)
 
+
+    def test_rejects_non_utf8_instruction_files(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp_dir:
+            root = Path(tmp_dir)
+            original_bytes = b"# Agent instructions\n\n\xff\xfe\n"
+            agents_file = root / "AGENTS.md"
+            agents_file.write_bytes(original_bytes)
+
+            with self.assertRaisesRegex(ValueError, "not valid UTF-8"):
+                build_dedupe_report(root, discover_instruction_files(root))
+
+            self.assertEqual(agents_file.read_bytes(), original_bytes)
+
     def test_rejects_symlinked_instruction_files(self) -> None:
         with tempfile.TemporaryDirectory() as tmp_dir:
             root = Path(tmp_dir)