pulseed 0.4.3 → 0.4.4

package/dist/orchestrator/goal/__tests__/goal-negotiator-decompose.test.js

@@ -1,876 +0,0 @@
-import { describe, it, expect, beforeEach, afterEach, vi } from "vitest";
-import * as fs from "node:fs";
-import { StateManager } from "../../../base/state/state-manager.js";
-import { EthicsGate } from "../../../platform/traits/ethics-gate.js";
-import { ObservationEngine } from "../../../platform/observation/observation-engine.js";
-import { GoalNegotiator, EthicsRejectedError } from "../goal-negotiator.js";
-import { GoalSchema } from "../../../base/types/goal.js";
-import { createMockLLMClient } from "../../../../tests/helpers/mock-llm.js";
-import { makeTempDir } from "../../../../tests/helpers/temp-dir.js";
-import { PASS_VERDICT_SAFE_JSON as PASS_VERDICT, REJECT_VERDICT_ILLEGAL_JSON as REJECT_VERDICT, } from "../../../../tests/helpers/ethics-fixtures.js";
-const SINGLE_DIMENSION_RESPONSE = JSON.stringify([
-    {
-        name: "completion_rate",
-        label: "Completion Rate",
-        threshold_type: "min",
-        threshold_value: 100,
-        observation_method_hint: "Check task completion metrics",
-    },
-]);
-const DIMENSIONS_RESPONSE = JSON.stringify([
-    {
-        name: "test_coverage",
-        label: "Test Coverage",
-        threshold_type: "min",
-        threshold_value: 80,
-        observation_method_hint: "Run test suite and check coverage report",
-    },
-    {
-        name: "code_quality",
-        label: "Code Quality Score",
-        threshold_type: "min",
-        threshold_value: 90,
-        observation_method_hint: "Run linter and code analysis tools",
-    },
-]);
-const FEASIBILITY_REALISTIC = JSON.stringify({
-    assessment: "realistic",
-    confidence: "high",
-    reasoning: "This target is achievable within the time horizon.",
-    key_assumptions: ["Current pace maintained"],
-    main_risks: [],
-});
-const FEASIBILITY_INFEASIBLE = JSON.stringify({
-    assessment: "infeasible",
-    confidence: "low",
-    reasoning: "This target is not achievable in the given timeframe.",
-    key_assumptions: ["No acceleration possible"],
-    main_risks: ["Target unreachable", "Burnout risk"],
-});
-const RESPONSE_MESSAGE_ACCEPT = "Your goal has been accepted. Let's get started!";
-const RESPONSE_MESSAGE_COUNTER = "This goal is too ambitious. Consider a safer target.";
-const SUBGOALS_RESPONSE = JSON.stringify([
-    {
-        title: "Setup CI Pipeline",
-        description: "Configure continuous integration for automated testing",
-        dimensions: [
-            {
-                name: "ci_configured",
-                label: "CI Configured",
-                threshold_type: "present",
-                threshold_value: null,
-                observation_method_hint: "Check CI config exists",
-            },
-        ],
-    },
-    {
-        title: "Write Unit Tests",
-        description: "Achieve 80% unit test coverage",
-        dimensions: [
-            {
-                name: "unit_coverage",
-                label: "Unit Test Coverage",
-                threshold_type: "min",
-                threshold_value: 80,
-                observation_method_hint: "Run coverage tool",
-            },
-        ],
-    },
-]);
-function makeTestGoal(overrides = {}) {
-    const now = new Date().toISOString();
-    return GoalSchema.parse({
-        id: overrides.id ?? "test-goal-id",
-        parent_id: null,
-        node_type: "goal",
-        title: overrides.title ?? "Test Goal",
-        description: overrides.description ?? "A test goal for testing",
-        status: "active",
-        dimensions: overrides.dimensions ?? [
-            {
-                name: "metric_a",
-                label: "Metric A",
-                current_value: 50,
-                threshold: { type: "min", value: 80 },
-                confidence: 0.8,
-                observation_method: {
-                    type: "mechanical",
-                    source: "test",
-                    schedule: null,
-                    endpoint: null,
-                    confidence_tier: "mechanical",
-                },
-                last_updated: now,
-                history: [
-                    { value: 30, timestamp: new Date(Date.now() - 30 * 24 * 60 * 60 * 1000).toISOString(), confidence: 0.8, source_observation_id: "obs-1" },
-                    { value: 50, timestamp: now, confidence: 0.8, source_observation_id: "obs-2" },
-                ],
-                weight: 1.0,
-                uncertainty_weight: null,
-                state_integrity: "ok",
-            },
-        ],
-        gap_aggregation: "max",
-        dimension_mapping: null,
-        constraints: ["No weekend work"],
-        children_ids: [],
-        target_date: null,
-        origin: "negotiation",
-        pace_snapshot: null,
-        deadline: null,
-        confidence_flag: "high",
-        user_override: false,
-        feasibility_note: null,
-        uncertainty_weight: 1.0,
-        created_at: now,
-        updated_at: now,
-    });
-}
-describe("GoalNegotiator", () => {
-    let tmpDir;
-    let stateManager;
-    let observationEngine;
-    beforeEach(() => {
-        tmpDir = makeTempDir();
-        stateManager = new StateManager(tmpDir);
-        observationEngine = new ObservationEngine(stateManager);
-    });
-    afterEach(() => {
-        fs.rmSync(tmpDir, { recursive: true, force: true, maxRetries: 3, retryDelay: 100 });
-    });
-    // ─── decompose() ───
-    describe("decompose()", () => {
-        it("returns subgoals and rejected subgoals", async () => {
-            const mockLLM = createMockLLMClient([
-                // For negotiate
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                // For decompose: LLM generates subgoals
-                SUBGOALS_RESPONSE,
-                // Ethics checks for each subgoal (2 subgoals)
-                PASS_VERDICT,
-                PASS_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const negotiateResult = await negotiator.negotiate("Improve quality");
-            const result = await negotiator.decompose(negotiateResult.goal.id, negotiateResult.goal);
-            expect(result.subgoals).toHaveLength(2);
-            expect(result.rejectedSubgoals).toHaveLength(0);
-        });
-        it("subgoals have correct parent_id", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                SUBGOALS_RESPONSE,
-                PASS_VERDICT,
-                PASS_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const negotiateResult = await negotiator.negotiate("Improve quality");
-            const result = await negotiator.decompose(negotiateResult.goal.id, negotiateResult.goal);
-            for (const subgoal of result.subgoals) {
-                expect(subgoal.parent_id).toBe(negotiateResult.goal.id);
-            }
-        });
-        it("subgoals have node_type subgoal", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                SUBGOALS_RESPONSE,
-                PASS_VERDICT,
-                PASS_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const negotiateResult = await negotiator.negotiate("Improve quality");
-            const result = await negotiator.decompose(negotiateResult.goal.id, negotiateResult.goal);
-            for (const subgoal of result.subgoals) {
-                expect(subgoal.node_type).toBe("subgoal");
-            }
-        });
-        it("subgoals have origin decomposition", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                SUBGOALS_RESPONSE,
-                PASS_VERDICT,
-                PASS_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const negotiateResult = await negotiator.negotiate("Improve quality");
-            const result = await negotiator.decompose(negotiateResult.goal.id, negotiateResult.goal);
-            for (const subgoal of result.subgoals) {
-                expect(subgoal.origin).toBe("decomposition");
-            }
-        });
-        it("rejected subgoals are collected with reason", async () => {
-            const mockLLM = createMockLLMClient([
-                // For negotiate
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                // For decompose
-                SUBGOALS_RESPONSE,
-                // First subgoal passes
-                PASS_VERDICT,
-                // Second subgoal rejected
-                REJECT_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const negotiateResult = await negotiator.negotiate("Improve quality");
-            const result = await negotiator.decompose(negotiateResult.goal.id, negotiateResult.goal);
-            expect(result.subgoals).toHaveLength(1);
-            expect(result.rejectedSubgoals).toHaveLength(1);
-            expect(result.rejectedSubgoals[0].reason).toBeTruthy();
-        });
-        it("persists accepted subgoals to state manager", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                SUBGOALS_RESPONSE,
-                PASS_VERDICT,
-                PASS_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const negotiateResult = await negotiator.negotiate("Improve quality");
-            const result = await negotiator.decompose(negotiateResult.goal.id, negotiateResult.goal);
-            for (const subgoal of result.subgoals) {
-                const loaded = await stateManager.loadGoal(subgoal.id);
-                expect(loaded).not.toBeNull();
-            }
-        });
-        it("all subgoals rejected returns empty subgoals array", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                SUBGOALS_RESPONSE,
-                REJECT_VERDICT,
-                REJECT_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const negotiateResult = await negotiator.negotiate("Improve quality");
-            const result = await negotiator.decompose(negotiateResult.goal.id, negotiateResult.goal);
-            expect(result.subgoals).toHaveLength(0);
-            expect(result.rejectedSubgoals).toHaveLength(2);
-        });
-    });
-    describe("decomposeIntoSubgoals()", () => {
-        it("returns null when GoalTreeManager is not provided", async () => {
-            const mockLLM = createMockLLMClient([]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            await expect(negotiator.decomposeIntoSubgoals("missing-goal")).resolves.toBeNull();
-        });
-        it("returns null when the goal does not exist", async () => {
-            const mockLLM = createMockLLMClient([]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const goalTreeManager = {
-                decomposeGoal: vi.fn(),
-            };
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine, undefined, undefined, goalTreeManager);
-            await expect(negotiator.decomposeIntoSubgoals("missing-goal")).resolves.toBeNull();
-            expect(goalTreeManager.decomposeGoal).not.toHaveBeenCalled();
-        });
-        it("uses default decomposition config when none is provided", async () => {
-            const parentGoal = makeTestGoal({ id: "goal-for-default-decompose" });
-            await stateManager.saveGoal(parentGoal);
-            const mockLLM = createMockLLMClient([]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const decomposeGoal = vi.fn().mockResolvedValue({
-                created: [],
-                reused: [],
-                pruned: [],
-                depth_reached: 0,
-            });
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine, undefined, undefined, { decomposeGoal });
-            await negotiator.decomposeIntoSubgoals(parentGoal.id);
-            expect(decomposeGoal).toHaveBeenCalledWith(parentGoal.id, {
-                max_depth: 5,
-                min_specificity: 0.7,
-                auto_prune_threshold: 0.3,
-                parallel_loop_limit: 3,
-            });
-        });
-        it("passes through an explicit decomposition config", async () => {
-            const parentGoal = makeTestGoal({ id: "goal-for-custom-decompose" });
-            await stateManager.saveGoal(parentGoal);
-            const mockLLM = createMockLLMClient([]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const decomposeGoal = vi.fn().mockResolvedValue({
-                created: [],
-                reused: [],
-                pruned: [],
-                depth_reached: 1,
-            });
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine, undefined, undefined, { decomposeGoal });
-            const config = {
-                max_depth: 2,
-                min_specificity: 0.9,
-                auto_prune_threshold: 0.2,
-                parallel_loop_limit: 1,
-            };
-            await negotiator.decomposeIntoSubgoals(parentGoal.id, config);
-            expect(decomposeGoal).toHaveBeenCalledWith(parentGoal.id, config);
-        });
-    });
-    describe("suggestGoals()", () => {
-        it("returns an empty array for blank context", async () => {
-            const mockLLM = createMockLLMClient([]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            await expect(negotiator.suggestGoals("   ")).resolves.toEqual([]);
-        });
-        it("returns parsed suggestions without persisting any goal", async () => {
-            const suggestionsResponse = JSON.stringify([
-                {
-                    title: "Increase Test Coverage",
-                    description: "Raise unit test coverage to 80 percent",
-                    rationale: "Coverage is below the team standard",
-                    dimensions_hint: ["test_coverage", "failing_tests"],
-                },
-            ]);
-            const mockLLM = createMockLLMClient([
-                suggestionsResponse,
-                PASS_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const suggestions = await negotiator.suggestGoals("The test suite lacks coverage", {
-                maxSuggestions: 1,
-            });
-            expect(suggestions).toHaveLength(1);
-            expect(suggestions[0]?.title).toBe("Increase Test Coverage");
-            expect(await stateManager.listGoalIds()).toEqual([]);
-        });
-    });
-    // ─── renegotiate() ───
-    describe("renegotiate()", () => {
-        it("throws when goal does not exist", async () => {
-            const mockLLM = createMockLLMClient([]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            await expect(negotiator.renegotiate("nonexistent", "user_request")).rejects.toThrow('goal "nonexistent" not found');
-        });
-        it("performs renegotiation on existing goal", async () => {
-            const mockLLM = createMockLLMClient([
-                // Initial negotiate
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                // Renegotiate
-                PASS_VERDICT, // ethics re-check
-                SINGLE_DIMENSION_RESPONSE, // re-decompose
-                FEASIBILITY_REALISTIC, // feasibility
-                RESPONSE_MESSAGE_ACCEPT, // response
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            const result = await negotiator.renegotiate(initial.goal.id, "user_request", "Need adjustments");
-            expect(result.goal).toBeDefined();
-            expect(result.response).toBeDefined();
-            expect(result.log).toBeDefined();
-        });
-        it("marks log as renegotiation", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            const result = await negotiator.renegotiate(initial.goal.id, "stall");
-            expect(result.log.is_renegotiation).toBe(true);
-            expect(result.log.renegotiation_trigger).toBe("stall");
-        });
-        it("records stall trigger correctly", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            const result = await negotiator.renegotiate(initial.goal.id, "stall");
-            expect(result.log.renegotiation_trigger).toBe("stall");
-        });
-        it("records new_info trigger correctly", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            const result = await negotiator.renegotiate(initial.goal.id, "new_info", "New data arrived");
-            expect(result.log.renegotiation_trigger).toBe("new_info");
-        });
-        it("records user_request trigger correctly", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            const result = await negotiator.renegotiate(initial.goal.id, "user_request");
-            expect(result.log.renegotiation_trigger).toBe("user_request");
-        });
-        it("throws EthicsRejectedError if re-check rejects", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                // renegotiate ethics check rejects
-                REJECT_VERDICT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            await expect(negotiator.renegotiate(initial.goal.id, "user_request")).rejects.toThrow(EthicsRejectedError);
-        });
-        it("updates the persisted negotiation log", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            await negotiator.renegotiate(initial.goal.id, "stall");
-            const log = await negotiator.getNegotiationLog(initial.goal.id);
-            expect(log).not.toBeNull();
-            expect(log.is_renegotiation).toBe(true);
-        });
-        it("updates goal in state manager", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const initial = await negotiator.negotiate("Test goal");
-            const renegResult = await negotiator.renegotiate(initial.goal.id, "user_request");
-            const loaded = await stateManager.loadGoal(initial.goal.id);
-            expect(loaded).not.toBeNull();
-            expect(loaded.updated_at).toBe(renegResult.goal.updated_at);
-        });
-    });
-    // ─── Negotiation log structure validation ───
-    describe("negotiation log structure", () => {
-        it("log has goal_id matching the goal", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.log.goal_id).toBe(result.goal.id);
-        });
-        it("log has valid timestamp", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(() => new Date(result.log.timestamp)).not.toThrow();
-        });
-        it("log renegotiation_trigger is null for initial negotiation", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.log.renegotiation_trigger).toBeNull();
-        });
-        it("counter_propose log includes counter_proposal details", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_INFEASIBLE,
-                RESPONSE_MESSAGE_COUNTER,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Unrealistic goal");
-            expect(result.log.step5_response.counter_proposal).not.toBeNull();
-            expect(typeof result.log.step5_response.counter_proposal.realistic_target).toBe("number");
-        });
-    });
-    // ─── Goal dimension thresholds ───
-    describe("goal dimension construction", () => {
-        it("min threshold creates correct structure", async () => {
-            const minDimension = JSON.stringify([
-                {
-                    name: "score",
-                    label: "Score",
-                    threshold_type: "min",
-                    threshold_value: 75,
-                    observation_method_hint: "Check score",
-                },
-            ]);
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                minDimension,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.goal.dimensions[0].threshold).toEqual({ type: "min", value: 75 });
-        });
-        it("present threshold creates correct structure", async () => {
-            const presentDimension = JSON.stringify([
-                {
-                    name: "feature_exists",
-                    label: "Feature Exists",
-                    threshold_type: "present",
-                    threshold_value: null,
-                    observation_method_hint: "Check if feature exists",
-                },
-            ]);
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                presentDimension,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.goal.dimensions[0].threshold).toEqual({ type: "present" });
-        });
-        it("match threshold creates correct structure", async () => {
-            const matchDimension = JSON.stringify([
-                {
-                    name: "status",
-                    label: "Status",
-                    threshold_type: "match",
-                    threshold_value: "deployed",
-                    observation_method_hint: "Check deployment status",
-                },
-            ]);
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                matchDimension,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.goal.dimensions[0].threshold).toEqual({ type: "match", value: "deployed" });
-        });
-        it("range threshold with array threshold_value creates correct structure", async () => {
-            const rangeDimension = JSON.stringify([
-                {
-                    name: "response_time",
-                    label: "Response Time",
-                    threshold_type: "range",
-                    threshold_value: [60, 80],
-                    observation_method_hint: "Measure response time in ms",
-                },
-            ]);
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                rangeDimension,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.goal.dimensions[0].threshold).toEqual({ type: "range", low: 60, high: 80 });
-        });
-        it("dimensions have initial null current_value", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.goal.dimensions[0].current_value).toBeNull();
-        });
-        it("dimensions have initial 0 confidence", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.goal.dimensions[0].confidence).toBe(0);
-        });
-    });
-    // ─── Feasibility with quantitative path (renegotiation with history) ───
-    describe("renegotiate() with quantitative feasibility path", () => {
-        it("uses quantitative path when history data is available", async () => {
-            // First, create a goal with history data
-            const goalWithHistory = makeTestGoal();
-            await stateManager.saveGoal(goalWithHistory);
-            // Dimension response that matches existing dimension name
-            const matchingDimension = JSON.stringify([
-                {
-                    name: "metric_a",
-                    label: "Metric A",
-                    threshold_type: "min",
-                    threshold_value: 80,
-                    observation_method_hint: "Measure metric A",
-                },
-            ]);
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT, // ethics re-check
-                matchingDimension, // re-decompose
-                // No qualitative feasibility call needed - quantitative path
-                RESPONSE_MESSAGE_ACCEPT, // response
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.renegotiate(goalWithHistory.id, "new_info");
-            expect(result.log.step4_evaluation).not.toBeNull();
-            // Should have used quantitative path since we have history
-            const dimEval = result.log.step4_evaluation.dimensions[0];
-            expect(dimEval.path).toBe("quantitative");
-            expect(dimEval.feasibility_ratio).not.toBeNull();
-        });
-    });
-    // ─── Multiple dimensions mixed feasibility ───
-    describe("negotiate() with mixed feasibility", () => {
-        it("counter_propose when any dimension is infeasible", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                DIMENSIONS_RESPONSE, // 2 dimensions
-                FEASIBILITY_REALISTIC, // dim 1 realistic
-                FEASIBILITY_INFEASIBLE, // dim 2 infeasible
-                RESPONSE_MESSAGE_COUNTER, // counter proposal
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Mixed feasibility goal");
-            expect(result.response.type).toBe("counter_propose");
-        });
-        it("accept when all dimensions are realistic", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                DIMENSIONS_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Realistic goal");
-            expect(result.response.type).toBe("accept");
-        });
-    });
-    // ─── Confidence flag on goal ───
-    describe("goal confidence_flag", () => {
-        it("sets high confidence when all feasibility is high", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_REALISTIC,
-                RESPONSE_MESSAGE_ACCEPT,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Test goal");
-            expect(result.goal.confidence_flag).toBe("high");
-        });
-        it("sets low confidence on infeasible dimensions", async () => {
-            const mockLLM = createMockLLMClient([
-                PASS_VERDICT,
-                SINGLE_DIMENSION_RESPONSE,
-                FEASIBILITY_INFEASIBLE,
-                RESPONSE_MESSAGE_COUNTER,
-            ]);
-            const ethicsGate = new EthicsGate(stateManager, mockLLM);
-            const negotiator = new GoalNegotiator(stateManager, mockLLM, ethicsGate, observationEngine);
-            const result = await negotiator.negotiate("Infeasible goal");
-            expect(result.goal.confidence_flag).toBe("low");
-        });
-    });
-    // ─── buildDecompositionPrompt — DataSource dimension placement ───
-    describe("buildDecompositionPrompt DataSource placement", () => {
-        it("DataSource dimensions appear before the example JSON when DataSources are available", async () => {
-            // We capture the prompt by inspecting what the LLM client receives.
-            let capturedPrompt = "";
-            let callCount = 0;
-            const capturingLLM = {
-                async sendMessage(messages, _options) {
-                    callCount++;
-                    if (callCount === 2) {
-                        // The decomposition call is the 2nd LLM call (after ethics check)
-                        capturedPrompt = messages[0]?.content ?? "";
-                    }
-                    // Return canned responses in order
-                    const responses = {
-                        1: PASS_VERDICT,
-                        2: SINGLE_DIMENSION_RESPONSE,
-                        3: FEASIBILITY_REALISTIC,
-                        4: RESPONSE_MESSAGE_ACCEPT,
-                    };
-                    const content = responses[callCount] ?? RESPONSE_MESSAGE_ACCEPT;
-                    return { content, usage: { input_tokens: 10, output_tokens: content.length }, stop_reason: "end_turn" };
-                },
-                parseJSON(content, schema) {
-                    return schema.parse(JSON.parse(content.trim()));
-                },
-            };
-            // Register a DataSource with known dimension names on the ObservationEngine
-            // We do this by calling registerDataSource directly via stateManager
-            // The simplest approach: use a mock ObservationEngine that returns a known dimension list
-            const mockObsEngine = {
-                getAvailableDimensionInfo() {
-                    return [{ name: "github_issues", dimensions: ["open_issue_count", "closed_issue_count", "completion_ratio"] }];
-                },
-            };
-            const ethicsGate = new EthicsGate(stateManager, capturingLLM);
-            const negotiator = new GoalNegotiator(stateManager, capturingLLM, ethicsGate, mockObsEngine);
-            await negotiator.negotiate("Improve project completion");
-            // The DataSource section should appear BEFORE the example JSON
-            const dsIndex = capturedPrompt.indexOf("DataSources");
-            const exampleIndex = capturedPrompt.indexOf("Example:");
-            expect(dsIndex).toBeGreaterThanOrEqual(0);
-            expect(exampleIndex).toBeGreaterThan(dsIndex);
-        });
-        it("DataSource CRITICAL CONSTRAINT text appears in prompt when DataSources are available", async () => {
-            let capturedPrompt = "";
-            let callCount = 0;
-            const capturingLLM = {
-                async sendMessage(messages, _options) {
-                    callCount++;
-                    if (callCount === 2) {
-                        capturedPrompt = messages[0]?.content ?? "";
-                    }
-                    const responses = {
-                        1: PASS_VERDICT,
-                        2: SINGLE_DIMENSION_RESPONSE,
-                        3: FEASIBILITY_REALISTIC,
-                        4: RESPONSE_MESSAGE_ACCEPT,
-                    };
-                    const content = responses[callCount] ?? RESPONSE_MESSAGE_ACCEPT;
-                    return { content, usage: { input_tokens: 10, output_tokens: content.length }, stop_reason: "end_turn" };
-                },
-                parseJSON(content, schema) {
-                    return schema.parse(JSON.parse(content.trim()));
-                },
-            };
-            const mockObsEngine = {
-                getAvailableDimensionInfo() {
-                    return [{ name: "sensor_feed", dimensions: ["temperature_celsius", "humidity_percent"] }];
-                },
-            };
-            const ethicsGate = new EthicsGate(stateManager, capturingLLM);
-            const negotiator = new GoalNegotiator(stateManager, capturingLLM, ethicsGate, mockObsEngine);
-            await negotiator.negotiate("Maintain comfortable environment");
-            expect(capturedPrompt).toContain("DataSources");
-            expect(capturedPrompt).toContain("use exact dimension names");
-            expect(capturedPrompt).toContain("temperature_celsius");
-            expect(capturedPrompt).toContain("humidity_percent");
-        });
-        it("no DataSource section in prompt when no DataSources are registered", async () => {
-            let capturedPrompt = "";
-            let callCount = 0;
-            const capturingLLM = {
-                async sendMessage(messages, _options) {
-                    callCount++;
-                    if (callCount === 2) {
-                        capturedPrompt = messages[0]?.content ?? "";
-                    }
-                    const responses = {
-                        1: PASS_VERDICT,
-                        2: SINGLE_DIMENSION_RESPONSE,
-                        3: FEASIBILITY_REALISTIC,
-                        4: RESPONSE_MESSAGE_ACCEPT,
-                    };
-                    const content = responses[callCount] ?? RESPONSE_MESSAGE_ACCEPT;
-                    return { content, usage: { input_tokens: 10, output_tokens: content.length }, stop_reason: "end_turn" };
-                },
-                parseJSON(content, schema) {
-                    return schema.parse(JSON.parse(content.trim()));
-                },
-            };
-            // observationEngine has no registered DataSources — getAvailableDimensionInfo() returns []
-            const ethicsGate = new EthicsGate(stateManager, capturingLLM);
-            const negotiator = new GoalNegotiator(stateManager, capturingLLM, ethicsGate, observationEngine);
-            await negotiator.negotiate("Simple goal with no data sources");
-            expect(capturedPrompt).not.toContain("CRITICAL CONSTRAINT");
-        });
-    });
-});
-//# sourceMappingURL=goal-negotiator-decompose.test.js.map