Route capability entries as evidence pool

tests/test_capability_evidence_pool.py

@@ -0,0 +1,72 @@
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from scripts.capability_evidence_pool import build_capability_evidence_pool_summary
+
+
+ROOT = Path(__file__).resolve().parents[1]
+REGISTRY = ROOT / "references" / "technique_registry.json"
+README = ROOT / "README.md"
+
+
+def test_registry_is_backend_evidence_pool_not_flat_user_skill_list() -> None:
+    registry = json.loads(REGISTRY.read_text(encoding="utf-8"))
+    techniques = registry["techniques"]
+
+    assert registry["registry_role"] == "backend_capability_evidence_pool"
+    assert registry["public_label"] == "89 capability entries"
+    assert "question-domain router" in registry["ordinary_user_policy"]
+
+    allowed_entry_types = {
+        "core_technique",
+        "supporting_indicator",
+        "composite_adjudicator",
+        "workflow_or_engineering",
+        "alias_entry",
+    }
+    allowed_roles = {"primary", "secondary", "context", "audit_only", "alias"}
+    allowed_visibility = {"ordinary_topic_router", "expert_audit", "hidden"}
+    allowed_prediction = {
+        "case_validated_partial",
+        "support_only",
+        "not_claimed",
+        "not_applicable",
+    }
+
+    for tech_id, tech in techniques.items():
+        assert tech["entry_type"] in allowed_entry_types, tech_id
+        assert tech["evidence_role"] in allowed_roles, tech_id
+        assert tech["user_visibility"] in allowed_visibility, tech_id
+        assert tech["verification_level"]["calculation"] in {"verified", "partial", "not_applicable"}, tech_id
+        assert tech["verification_level"]["rule"] in {"verified", "partial", "not_applicable"}, tech_id
+        assert tech["verification_level"]["prediction"] in allowed_prediction, tech_id
+        assert tech["conclusion_policy"], tech_id
+
+    assert techniques["case_validator"]["evidence_role"] == "audit_only"
+    assert techniques["thematic_report_orchestrator"]["entry_type"] == "workflow_or_engineering"
+    assert techniques["neechabhanga"]["evidence_role"] == "alias"
+    assert techniques["special_lagnas"]["evidence_role"] == "alias"
+
+
+def test_evidence_pool_summary_routes_few_primary_items_and_many_support_items() -> None:
+    summary = build_capability_evidence_pool_summary()
+
+    assert summary["scope"] == "backend_capability_evidence_pool"
+    assert summary["total_entries"] == 89
+    assert summary["ordinary_user_policy"].startswith("Users see topic-level")
+    assert summary["evidence_role_counts"]["primary"] >= 8
+    assert summary["evidence_role_counts"]["secondary"] > summary["evidence_role_counts"]["primary"]
+    assert summary["evidence_role_counts"]["audit_only"] >= 3
+    assert summary["prediction_verification_counts"]["not_claimed"] > 0
+    assert summary["conclusion_policy"]["primary_chain_required"] is True
+    assert summary["conclusion_policy"]["all_89_entries_must_not_be_flattened_into_conclusions"] is True
+
+
+def test_readme_uses_capability_entries_language_instead_of_89_techniques_claim() -> None:
+    readme = README.read_text(encoding="utf-8")
+
+    assert "89 capability entries" in readme
+    assert "89 techniques" not in readme
+    assert "backend evidence pool" in readme

tests/test_cli_smoke.py

@@ -334,6 +334,11 @@ def test_full_reading_reports_ayanamsa_metadata_and_ai_prompt_pack() -> None:
     assert vedastro_rows[0]["status"] in {"used", "blocked"}
     assert "overview only" in vedastro_rows[0]["note"]
     assert "domain_statuses" in vedastro_rows[0]["note"]
+    capability_pool = prompt_pack["evidence_snapshot"]["capability_evidence_pool"]
+    assert capability_pool["scope"] == "backend_capability_evidence_pool"
+    assert capability_pool["total_entries"] == 89
+    assert capability_pool["conclusion_policy"]["all_89_entries_must_not_be_flattened_into_conclusions"] is True
+    assert "后台备选证据池" in prompt_pack["prompt_zh"]
 
 
 def test_full_reading_generates_guided_topics_from_real_evidence() -> None:

tests/test_readme_badges.py

@@ -27,7 +27,7 @@ def test_readme_badges_match_technique_registry_counts() -> None:
     counts = Counter(item["status"] for item in techniques)
     total = len(registry["techniques"])
 
-    assert _readme_badge_value("Techniques") == total
+    assert _readme_badge_value("Capabilities") == total
     assert _readme_badge_value("Covered") == counts["covered"]
     assert _readme_badge_value("Complete") == counts["complete"]
     assert _readme_badge_value("Partial") == counts["partial"]