feat(experiment): checkpoint focused investigation boundaries

2026-08-19 05:41:32 +01:00
parent c0b963973f
commit 7add85d8d2
4 changed files with 466 additions and 0 deletions
@@ -0,0 +1,89 @@
+import { getProvider } from "@/lib/llm/provider";
+import { buildFocusedDeconstructPrompt, validateFocusedDeconstructSchema } from "@/lib/graph/focused-investigation";
+
+export async function POST(request) {
+  try {
+    const body = await request.json();
+
+    if (!body.targetNodeId || typeof body.targetNodeId !== "string") {
+      return Response.json(
+        { error: "Request must include a 'targetNodeId' string field" },
+        { status: 400 },
+      );
+    }
+    if (!body.targetLabel || typeof body.targetLabel !== "string") {
+      return Response.json(
+        { error: "Request must include a 'targetLabel' string field" },
+        { status: 400 },
+      );
+    }
+    if (!body.targetDescription || typeof body.targetDescription !== "string") {
+      return Response.json(
+        { error: "Request must include a 'targetDescription' string field" },
+        { status: 400 },
+      );
+    }
+    if (!body.centralStatement || typeof body.centralStatement !== "string") {
+      return Response.json(
+        { error: "Request must include a 'centralStatement' string field" },
+        { status: 400 },
+      );
+    }
+    if (!body.question || typeof body.question !== "string") {
+      return Response.json(
+        { error: "Request must include a 'question' string field" },
+        { status: 400 },
+      );
+    }
+    if (!body.answer || typeof body.answer !== "string") {
+      return Response.json(
+        { error: "Request must include an 'answer' string field" },
+        { status: 400 },
+      );
+    }
+
+    const prompt = buildFocusedDeconstructPrompt({
+      targetLabel: body.targetLabel,
+      targetDescription: body.targetDescription,
+      centralStatement: body.centralStatement,
+      question: body.question,
+      answer: body.answer,
+    });
+
+    const provider = getProvider();
+    const startedAt = Date.now();
+    const raw = await provider.generateReconstruction(prompt, process.env.OLLAMA_MODEL);
+    const elapsedMs = Date.now() - startedAt;
+
+    // Validate schema (required fields present, no graph-mutation fields)
+    const validationErrors = validateFocusedDeconstructSchema(raw);
+    if (validationErrors.length > 0) {
+      return Response.json(
+        {
+          success: false,
+          error: "Focused deconstruction result did not match expected schema",
+          validationErrors,
+          targetNodeId: body.targetNodeId,
+          elapsedMs,
+        },
+        { status: 502 },
+      );
+    }
+
+    return Response.json({
+      success: true,
+      targetNodeId: raw.targetNodeId,
+      observations: raw.observations,
+      uncertainties: raw.uncertainties,
+      assumptions: raw.assumptions,
+      relationships: raw.relationships,
+      possibleFollowUpQuestions: raw.possibleFollowUpQuestions,
+      elapsedMs,
+    });
+  } catch (e) {
+    return Response.json(
+      { error: e.message || "Unknown server error" },
+      { status: 500 },
+    );
+  }
+}
@@ -0,0 +1,52 @@
+import { formulateQuestionForTarget } from "@/lib/graph/focused-investigation";
+
+export async function POST(request) {
+  try {
+    const body = await request.json();
+
+    if (!body.targetNodeId || typeof body.targetNodeId !== "string") {
+      return Response.json(
+        { error: "Request must include a 'targetNodeId' string field" },
+        { status: 400 },
+      );
+    }
+
+    if (!body.situationGraph || typeof body.situationGraph !== "object") {
+      return Response.json(
+        { error: "Request must include a 'situationGraph' object field" },
+        { status: 400 },
+      );
+    }
+
+    const result = formulateQuestionForTarget({
+      situationGraph: body.situationGraph,
+      targetNodeId: body.targetNodeId,
+    });
+
+    if (!result.success) {
+      return Response.json(
+        { success: false, error: result.error },
+        { status: 400 },
+      );
+    }
+
+    return Response.json({
+      success: true,
+      targetNodeId: result.targetNodeId,
+      question: result.question,
+      strategy: result.strategy,
+      reasoningPattern: result.reasoningPattern,
+      reasoningPatternReason: result.reasoningPatternReason,
+      reason: result.reason,
+      questionFamily: result.questionFamily,
+      selectedQuestionTemplate: result.selectedQuestionTemplate,
+      allowedQuestionFamilies: result.allowedQuestionFamilies,
+      rejectedQuestionFamilies: result.rejectedQuestionFamilies,
+    });
+  } catch (e) {
+    return Response.json(
+      { error: e.message || "Unknown server error" },
+      { status: 500 },
+    );
+  }
+}
@@ -0,0 +1,139 @@
+import { formulateQuestion } from "@/lib/graph/question-formulator.js";
+
+const FOCUSED_ANSWER_SCHEMA_FIELDS = [
+  "targetNodeId",
+  "observations",
+  "uncertainties",
+  "assumptions",
+  "relationships",
+  "possibleFollowUpQuestions",
+];
+
+const FORBIDDEN_GRAPH_MUTATION_FIELDS = [
+  "addedNodes",
+  "updatedNodes",
+  "removedNodes",
+  "addedEdges",
+  "removedEdges",
+  "resolvedNodeIds",
+  "activeUnknownNodeId",
+  "selectedQuestion",
+];
+
+/**
+ * Formulate a question for the explicitly user-selected unresolved node.
+ *
+ * Input: { situationGraph, targetNodeId }
+ * Validates that target exists, is an unknown, and is not resolved.
+ * Calls formulateQuestion with only factual explicit-user-targeting context.
+ * Does NOT invoke selectActiveUnknownCandidate, determineGraphBackedQuestion,
+ * global ranking, or global recommendation.
+ * Does NOT mutate activeUnknownNodeId, selectedQuestion, SituationGraph, or resolvedNodeIds.
+ */
+export function formulateQuestionForTarget({ situationGraph, targetNodeId }) {
+  const nodesById = new Map(situationGraph?.nodes?.map((n) => [n.id, n]) || []);
+  const targetNode = nodesById.get(targetNodeId);
+
+  if (!targetNode) {
+    return { success: false, error: `Target node ${targetNodeId} not found in graph.` };
+  }
+
+  if (targetNode.kind !== "unknown") {
+    return { success: false, error: `Target node ${targetNodeId} is not an unknown (kind=${targetNode.kind}).` };
+  }
+
+  if (targetNode.status === "resolved") {
+    return { success: false, error: `Target node ${targetNodeId} is already resolved.` };
+  }
+
+  // Build explicit user-targeting context — only factual fields from the targeting request.
+  // No activeUnknownNodeId override, no selectedQuestion mutation, no global selection state.
+  const context = {
+    resolvedValues: [],
+    suppressDecisionSufficiencyConfirmation: false,
+    _explicitTargetNodeId: targetNodeId,
+    _explicitTargetLabel: targetNode.label,
+    _explicitTargetDescription: targetNode.description,
+  };
+
+  const result = formulateQuestion({ node: targetNode, graph: situationGraph, context });
+
+  return {
+    success: true,
+    targetNodeId,
+    question: result.question,
+    strategy: result.strategy ?? null,
+    reasoningPattern: result.reasoningPattern,
+    reasoningPatternReason: result.reasoningPatternReason,
+    reason: result.reason,
+    questionFamily: result.questionFamily,
+    selectedQuestionTemplate: result.selectedQuestionTemplate,
+    allowedQuestionFamilies: result.allowedQuestionFamilies,
+    rejectedQuestionFamilies: result.rejectedQuestionFamilies,
+  };
+}
+
+/**
+ * Build a focused answer deconstruction prompt for one explicitly user-chosen investigation.
+ */
+export function buildFocusedDeconstructPrompt({ targetLabel, targetDescription, centralStatement, question, answer }) {
+  return `You are performing focused answer deconstruction for one explicitly user-chosen investigation.
+
+Return exactly one JSON object. Return JSON only.
+
+This is NOT a graph update task.
+Do NOT output graph mutations.
+Do NOT output selection, ranking, ownership, recommendation, confidence, or next-best-question semantics.
+Do NOT include any of these fields: addedNodes, updatedNodes, removedNodes, addedEdges, removedEdges, resolvedNodeIds, activeUnknownNodeId, selectedQuestion.
+
+Required top-level fields:
+- targetNodeId
+- observations
+- uncertainties
+- assumptions
+- relationships
+- possibleFollowUpQuestions
+
+Field rules:
+- targetNodeId must be included as a string identifying this investigation node
+- observations: only statements directly supported by the answer
+- uncertainties: only things the answer explicitly leaves unknown or unclear
+- assumptions: include only if the answer itself relies on an assumption
+- relationships: only direct supported relationships among extracted items, each with { from, to, type, rationale }
+- possibleFollowUpQuestions: unresolved questions genuinely exposed by this answer, unranked
+
+Focused case context:
+- target label: ${targetLabel}
+- target description: ${targetDescription}
+- central case statement: ${centralStatement}
+
+Question:
+${question}
+
+Answer:
+${answer}`;
+}
+
+/**
+ * Validate that a focused-answer deconstruction result contains only the expected fields.
+ * Returns an array of errors (empty = valid).
+ */
+export function validateFocusedDeconstructSchema(result) {
+  const errors = [];
+
+  for (const field of FOCUSED_ANSWER_SCHEMA_FIELDS) {
+    if (!(field in result)) {
+      errors.push(`Missing required field: ${field}`);
+    }
+  }
+
+  for (const field of FORBIDDEN_GRAPH_MUTATION_FIELDS) {
+    if (field in result) {
+      errors.push(`Forbidden graph-mutation field present: ${field}`);
+    }
+  }
+
+  return errors;
+}
+
+export { FOCUSED_ANSWER_SCHEMA_FIELDS, FORBIDDEN_GRAPH_MUTATION_FIELDS };
@@ -0,0 +1,186 @@
+import { describe, expect, it } from "vitest";
+import { makeNode, makeGraph } from "@/lib/graph/schema.js";
+import { formulateQuestionForTarget, buildFocusedDeconstructPrompt, validateFocusedDeconstructSchema } from "@/lib/graph/focused-investigation.js";
+
+// ── helpers ──────────────────────────────────────────────────────────────
+
+function makeTestGraph() {
+  return makeGraph({
+    centralStatement: "Should we launch the product now?",
+    currentSummary: "initial summary",
+    nodes: [
+      makeNode({ id: "a", label: "Market demand signal", description: "Evidence that customers want this.", kind: "unknown", status: "unknown" }),
+      makeNode({ id: "b", label: "Competitor activity", description: "What others are doing.", kind: "observation", status: "supported" }),
+    ],
+    edges: [],
+    resolvedNodeIds: ["b"],
+  });
+}
+
+// ── Boundary 1: explicit formulation ─────────────────────────────────────
+
+describe("formulateQuestionForTarget — explicit node selection", () => {
+  it("validates that the target node exists in the graph", () => {
+    const graph = makeTestGraph();
+    const result = formulateQuestionForTarget({ situationGraph: graph, targetNodeId: "nonexistent" });
+    expect(result.success).toBe(false);
+    expect(result.error).toContain("not found");
+  });
+
+  it("validates that the target node is an unknown", () => {
+    const graph = makeTestGraph();
+    const result = formulateQuestionForTarget({ situationGraph: graph, targetNodeId: "b" });
+    expect(result.success).toBe(false);
+    expect(result.error).toContain("not an unknown");
+  });
+
+  it("validates that the target node is not resolved", () => {
+    const graph = makeTestGraph();
+    // Create a resolved unknown node
+    graph.nodes.push(makeNode({ id: "c", label: "Done", description: "Completed item", kind: "unknown", status: "resolved" }));
+    const result = formulateQuestionForTarget({ situationGraph: graph, targetNodeId: "c" });
+    expect(result.success).toBe(false);
+    expect(result.error).toContain("resolved");
+  });
+
+  it("uses the supplied targetNodeId in the formulation output", () => {
+    const graph = makeTestGraph();
+    const result = formulateQuestionForTarget({ situationGraph: graph, targetNodeId: "a" });
+    expect(result.success).toBe(true);
+    expect(result.targetNodeId).toBe("a");
+  });
+
+  it("does not mutate activeUnknownNodeId on the graph", () => {
+    const graph = makeTestGraph();
+    const before = graph.activeUnknownNodeId;
+    formulateQuestionForTarget({ situationGraph: graph, targetNodeId: "a" });
+    expect(graph.activeUnknownNodeId).toBe(before);
+  });
+
+  it("does not mutate SituationGraph in any way", () => {
+    const graph = makeTestGraph();
+    const snapshot = JSON.stringify(graph);
+    formulateQuestionForTarget({ situationGraph: graph, targetNodeId: "a" });
+    expect(JSON.stringify(graph)).toBe(snapshot);
+  });
+
+  it("does not invoke global selection (selectActiveUnknownCandidate / determineGraphBackedQuestion)", () => {
+    const graph = makeTestGraph();
+    const result = formulateQuestionForTarget({ situationGraph: graph, targetNodeId: "a" });
+    expect(result.success).toBe(true);
+    expect(result.targetNodeId).toBe("a");
+    expect(result.question).toBeDefined();
+    expect(typeof result.question).toBe("string");
+    expect(result.question.length).toBeGreaterThan(0);
+  });
+});
+
+// ── Boundary 2: focused answer deconstruction ────────────────────────────
+
+describe("buildFocusedDeconstructPrompt", () => {
+  it("produces a prompt containing the target label", () => {
+    const prompt = buildFocusedDeconstructPrompt({
+      targetLabel: "Market demand signal",
+      targetDescription: "Evidence that customers want this.",
+      centralStatement: "Should we launch?",
+      question: "What evidence would clarify market demand?",
+      answer: "Some evidence suggests demand.",
+    });
+    expect(prompt).toContain("Market demand signal");
+  });
+
+  it("produces a prompt containing the answer text", () => {
+    const prompt = buildFocusedDeconstructPrompt({
+      targetLabel: "X",
+      targetDescription: "Y",
+      centralStatement: "Z",
+      question: "Q?",
+      answer: "Some evidence suggests demand.",
+    });
+    expect(prompt).toContain("Some evidence suggests demand.");
+  });
+
+  it("does NOT output graph-mutation fields (only lists them as forbidden in the instructions)", () => {
+    const prompt = buildFocusedDeconstructPrompt({
+      targetLabel: "X",
+      targetDescription: "Y",
+      centralStatement: "Z",
+      question: "Q?",
+      answer: "A.",
+    });
+    // The prompt instructs the model NOT to output these fields.
+    // It mentions them only as forbidden outputs, not as required outputs.
+    expect(prompt).toContain("Do NOT include any of these fields");
+    expect(prompt).not.toContain("- addedNodes");
+    expect(prompt).not.toContain("- updatedNodes");
+    expect(prompt).not.toContain("- removedNodes");
+    expect(prompt).not.toContain("- resolvedNodeIds");
+  });
+});
+
+describe("validateFocusedDeconstructSchema", () => {
+  it("passes when all required fields are present", () => {
+    const result = { targetNodeId: "a", observations: [], uncertainties: [], assumptions: [], relationships: [], possibleFollowUpQuestions: [] };
+    expect(validateFocusedDeconstructSchema(result)).toEqual([]);
+  });
+
+  it("fails when a required field is missing", () => {
+    const result = { targetNodeId: "a", observations: [], uncertainties: [], assumptions: [] };
+    const errors = validateFocusedDeconstructSchema(result);
+    expect(errors.length).toBeGreaterThan(0);
+    expect(errors.some((e) => e.includes("relationships") || e.includes("possibleFollowUpQuestions"))).toBeTruthy();
+  });
+
+  it("rejects results that contain graph-mutation fields", () => {
+    const result = {
+      targetNodeId: "a", observations: [], uncertainties: [], assumptions: [], relationships: [], possibleFollowUpQuestions: [],
+      addedNodes: [{ id: "x" }],
+    };
+    const errors = validateFocusedDeconstructSchema(result);
+    expect(errors.some((e) => e.includes("addedNodes"))).toBeTruthy();
+  });
+
+  it("rejects results that contain resolvedNodeIds", () => {
+    const result = {
+      targetNodeId: "a", observations: [], uncertainties: [], assumptions: [], relationships: [], possibleFollowUpQuestions: [],
+      resolvedNodeIds: ["x"],
+    };
+    const errors = validateFocusedDeconstructSchema(result);
+    expect(errors.some((e) => e.includes("resolvedNodeIds"))).toBeTruthy();
+  });
+
+  it("rejects results that contain activeUnknownNodeId", () => {
+    const result = {
+      targetNodeId: "a", observations: [], uncertainties: [], assumptions: [], relationships: [], possibleFollowUpQuestions: [],
+      activeUnknownNodeId: "x",
+    };
+    const errors = validateFocusedDeconstructSchema(result);
+    expect(errors.some((e) => e.includes("activeUnknownNodeId"))).toBeTruthy();
+  });
+
+  it("rejects results that contain selectedQuestion", () => {
+    const result = {
+      targetNodeId: "a", observations: [], uncertainties: [], assumptions: [], relationships: [], possibleFollowUpQuestions: [],
+      selectedQuestion: "some question",
+    };
+    const errors = validateFocusedDeconstructSchema(result);
+    expect(errors.some((e) => e.includes("selectedQuestion"))).toBeTruthy();
+  });
+
+  it("all required fields are observations, uncertainties, assumptions, relationships, possibleFollowUpQuestions", () => {
+    const fullResult = { targetNodeId: "a", observations: [], uncertainties: [], assumptions: [], relationships: [], possibleFollowUpQuestions: [] };
+    const errors = validateFocusedDeconstructSchema(fullResult);
+    expect(errors.length).toBe(0);
+  });
+
+  it("focused result contains no graph-mutation fields when valid", () => {
+    const fullResult = { targetNodeId: "a", observations: [], uncertainties: [], assumptions: [], relationships: [], possibleFollowUpQuestions: [] };
+    const errors = validateFocusedDeconstructSchema(fullResult);
+    expect(errors.length).toBe(0);
+    // Confirm no forbidden keys are in the result
+    const forbidden = ["addedNodes", "updatedNodes", "removedNodes", "addedEdges", "removedEdges", "resolvedNodeIds", "activeUnknownNodeId", "selectedQuestion"];
+    for (const key of forbidden) {
+      expect(key in fullResult).toBe(false);
+    }
+  });
+});