Add self-scoping assessment comparison

tests/test_cli.py

@@ -131,3 +131,26 @@ def test_export_assessment_cli_writes_completed_run_artifact(tmp_path):
     assert artifact["execution"]["analysis_run_id"] == summary.analysis_run.id
     assert artifact["assessment"]["role"] == "challenger"
     assert artifact["generated_tree"]["abilities"]
+
+
+def test_compare_assessment_cli_writes_markdown_report(tmp_path):
+    output_path = tmp_path / "comparison.md"
+
+    exit_code = main(
+        [
+            "compare-assessment",
+            "--golden",
+            "docs/self-scoping/golden/repo-scoping-golden-profile.v1.json",
+            "--assessment",
+            "docs/self-scoping/assessments/repo-scoping-known-bad-2026-05-15-run-39.json",
+            "--output",
+            str(output_path),
+            "--format",
+            "markdown",
+        ]
+    )
+
+    report = output_path.read_text(encoding="utf-8")
+    assert exit_code == 0
+    assert "Status: `regression`" in report
+    assert "Route LLM Requests Across Providers" in report

tests/test_self_scoping_comparison.py

@@ -0,0 +1,62 @@
+from pathlib import Path
+
+from repo_registry.self_scoping.comparison import (
+    compare_assessment_to_golden,
+    comparison_markdown,
+    load_json,
+)
+
+
+ROOT = Path(__file__).resolve().parents[1]
+GOLDEN_PROFILE = (
+    ROOT
+    / "docs"
+    / "self-scoping"
+    / "golden"
+    / "repo-scoping-golden-profile.v1.json"
+)
+KNOWN_BAD = (
+    ROOT
+    / "docs"
+    / "self-scoping"
+    / "assessments"
+    / "repo-scoping-known-bad-2026-05-15-run-39.json"
+)
+
+
+def test_compare_known_bad_assessment_to_golden_flags_regression():
+    comparison = compare_assessment_to_golden(
+        load_json(GOLDEN_PROFILE),
+        load_json(KNOWN_BAD),
+    )
+
+    assert comparison["schema_version"] == "self-scoping-comparison/v1"
+    assert comparison["status"] == "regression"
+    assert "Route LLM Requests Across Providers" in comparison[
+        "forbidden_native_capabilities_present"
+    ]
+    assert "Scan Repositories Into Observed Facts" in comparison[
+        "missing_expected_capabilities"
+    ]
+    assert {item["id"] for item in comparison["known_regression_patterns"]} >= {
+        "RREG-SELF-REG-001",
+        "RREG-SELF-REG-002",
+        "RREG-SELF-REG-003",
+    }
+    assert any(
+        item["feature_type"] == "API" for item in comparison["misplaced_features"]
+    )
+
+
+def test_comparison_markdown_summarizes_actionable_sections():
+    comparison = compare_assessment_to_golden(
+        load_json(GOLDEN_PROFILE),
+        load_json(KNOWN_BAD),
+    )
+
+    markdown = comparison_markdown(comparison)
+
+    assert "# Self-Scoping Comparison" in markdown
+    assert "## Missing Expected Capabilities" in markdown
+    assert "## Forbidden Native Capabilities Present" in markdown
+    assert "Route LLM Requests Across Providers" in markdown