Planner Autonomy: закрепить alignment guard для catalog chains

2026-05-01 14:47:19 +03:00
parent 417b51096e
commit 67f5e908c9
3 changed files with 151 additions and 2 deletions
@@ -63,6 +63,147 @@ describe("assistant MCP discovery planner", () => {
    expect(result.reason_codes).toContain("planner_catalog_chain_template_search_top_value_flow");
  });

+  it("keeps representative graph-selected chains aligned with top catalog template matches", () => {
+    const graph = (
+      businessFactFamily: string,
+      actionFamily: string,
+      extra: {
+        subject_candidates?: string[];
+        comparison_need?: string | null;
+        ranking_need?: string | null;
+      } = {}
+    ) => ({
+      schema_version: "assistant_data_need_graph_v1" as const,
+      policy_owner: "assistantMcpDiscoveryDataNeedGraph" as const,
+      subject_candidates: extra.subject_candidates ?? ["SVK"],
+      business_fact_family: businessFactFamily,
+      action_family: actionFamily,
+      aggregation_need: null,
+      time_scope_need: "explicit_period",
+      comparison_need: extra.comparison_need ?? null,
+      ranking_need: extra.ranking_need ?? null,
+      proof_expectation: "coverage_checked_fact",
+      clarification_gaps: [],
+      decomposition_candidates: [],
+      forbidden_overclaim_flags: ["no_raw_model_claims", "no_unchecked_fact_totals"],
+      reason_codes: ["data_need_graph_built"]
+    });
+    const cases = [
+      {
+        name: "value_flow",
+        expected: "value_flow",
+        input: {
+          dataNeedGraph: graph("value_flow", "turnover"),
+          turnMeaning: {
+            asked_action_family: "turnover",
+            explicit_entity_candidates: ["SVK"],
+            explicit_date_scope: "2020"
+          }
+        }
+      },
+      {
+        name: "value_flow_comparison",
+        expected: "value_flow_comparison",
+        input: {
+          dataNeedGraph: graph("value_flow", "net_value_flow", { comparison_need: "incoming_vs_outgoing" }),
+          turnMeaning: {
+            asked_action_family: "net_value_flow",
+            explicit_entity_candidates: ["SVK"],
+            explicit_date_scope: "2020"
+          }
+        }
+      },
+      {
+        name: "value_flow_ranking",
+        expected: "value_flow_ranking",
+        input: {
+          dataNeedGraph: graph("value_flow", "turnover", { subject_candidates: [], ranking_need: "top_desc" }),
+          turnMeaning: {
+            asked_action_family: "turnover",
+            explicit_date_scope: "2020",
+            explicit_organization_scope: "Org"
+          }
+        }
+      },
+      {
+        name: "document_evidence",
+        expected: "document_evidence",
+        input: {
+          dataNeedGraph: graph("document_evidence", "list_documents"),
+          turnMeaning: {
+            asked_action_family: "list_documents",
+            explicit_entity_candidates: ["SVK"],
+            explicit_date_scope: "2020"
+          }
+        }
+      },
+      {
+        name: "movement_evidence",
+        expected: "movement_evidence",
+        input: {
+          dataNeedGraph: graph("movement_evidence", "list_movements"),
+          turnMeaning: {
+            asked_action_family: "list_movements",
+            explicit_entity_candidates: ["SVK"],
+            explicit_date_scope: "2020"
+          }
+        }
+      },
+      {
+        name: "schema_surface",
+        expected: "metadata_inspection",
+        input: {
+          dataNeedGraph: graph("schema_surface", "inspect_catalog", { subject_candidates: [] }),
+          turnMeaning: {
+            asked_action_family: "inspect_catalog"
+          }
+        }
+      },
+      {
+        name: "entity_resolution",
+        expected: "entity_resolution",
+        input: {
+          dataNeedGraph: graph("entity_grounding", "search_business_entity"),
+          turnMeaning: {
+            asked_action_family: "search_business_entity",
+            explicit_entity_candidates: ["SVK"]
+          }
+        }
+      },
+      {
+        name: "lifecycle",
+        expected: "lifecycle",
+        input: {
+          dataNeedGraph: graph("activity_lifecycle", "activity_duration"),
+          turnMeaning: {
+            asked_action_family: "activity_duration",
+            explicit_entity_candidates: ["SVK"]
+          }
+        }
+      },
+      {
+        name: "inventory_stock_snapshot",
+        expected: "inventory_stock_snapshot",
+        input: {
+          dataNeedGraph: graph("inventory_stock_snapshot", "stock_snapshot", { subject_candidates: [] }),
+          turnMeaning: {
+            asked_action_family: "stock_snapshot",
+            explicit_date_scope: "2020",
+            explicit_organization_scope: "Org"
+          }
+        }
+      }
+    ] as const;
+
+    for (const item of cases) {
+      const result = planAssistantMcpDiscovery(item.input);
+      expect(result.selected_chain_id, item.name).toBe(item.expected);
+      expect(result.catalog_chain_template_alignment.top_chain_template_match, item.name).toBe(item.expected);
+      expect(result.catalog_chain_template_alignment.selected_chain_template_rank, item.name).toBe(1);
+      expect(result.catalog_chain_template_alignment.selected_chain_matches_top, item.name).toBe(true);
+    }
+  });
+
  it("keeps a value-flow plan in clarification state when period axis is missing", () => {
    const result = planAssistantMcpDiscovery({
      turnMeaning: {