diff --git a/lib/graph/prompt-builder.js b/lib/graph/prompt-builder.js
index b8896ae..dc11a9a 100644
--- a/lib/graph/prompt-builder.js
+++ b/lib/graph/prompt-builder.js
@@ -134,6 +134,19 @@ The JSON object must contain exactly these top-level fields:
 31. If the answer explicitly states a hard constraint, state that directly in userSupportedMeaning.
 32. Populate resolutionGuidance when the user's meaning genuinely implies must_remain_unresolved, may_resolve, or must_resolve. Keep it null only when no existing resolution state actually applies.
 
+## Decision Option Structure Rules
+When the user presents mutually exclusive candidate actions for one unresolved choice:
+
+1. Create exactly one node of kind "unknown" to carry the decision question (the existing mechanism). Do not add a separate "decision" node kind. Keep that unknown as-is or create it fresh — do not duplicate it into every option.
+
+2. For each candidate path, create exactly one node of kind "option". The option's label names the alternative; its description states what that alternative entails.
+
+3. Link each option to the decision-context unknown using relationship "contained_in" (edge: option → unknown). Shared membership already implies these options are alternatives of each other — do not add an "alternative_to" edge between options.
+
+4. Attach consequences and evidence to the specific option they belong to via existing edge types ("causes", "may_cause", etc.). Each consequence's fromNodeId explicitly identifies its parent option. Do not collapse all alternatives into one generic trade-off description on a single node.
+
+5. A do-nothing / stay-put / current-state path is an option when it is genuinely one of the alternatives — represent it with kind "option" and label it clearly. Do not introduce an "is_baseline", "is_default", or "is_status_quo" field; baseline meaning is carried by label and consequences alone in this implementation.
+
 ## Contract: structuralActionRequired Declaration Rule
 
 When answerMeaning.userSupportedMeaning is populated you MUST set structuralActionRequired to match what your proposal outputs:
diff --git a/lib/graph/schema.js b/lib/graph/schema.js
index b2bf340..35eb491 100644
--- a/lib/graph/schema.js
+++ b/lib/graph/schema.js
@@ -18,6 +18,7 @@ export const SituationKind = /** @type {const} */ ({
   assumption: "assumption",
   unknown: "unknown",
   conclusion: "conclusion",
+  option: "option",
 });
 
 export const SituationStatus = /** @type {const} */ ({
@@ -83,6 +84,7 @@ export const SituationRelationship = /** @type {const} */ ({
   measures: "measures",
   compares_with: "compares_with",
   updates: "updates",
+  contained_in: "contained_in",
   other: "other",
 });
 
diff --git a/tests/graph/prompt-builder.test.js b/tests/graph/prompt-builder.test.js
index a72efa2..e22d4c8 100644
--- a/tests/graph/prompt-builder.test.js
+++ b/tests/graph/prompt-builder.test.js
@@ -79,13 +79,13 @@ describe("buildGraphUpdatePrompt", () => {
   it("lists enum values", () => {
     const prompt = buildGraphUpdatePrompt(makeContext());
     expect(prompt).toContain(
-      "observation | reported_claim | metric | state | transition | relationship | assumption | unknown | conclusion",
+      "observation | reported_claim | metric | state | transition | relationship | assumption | unknown | conclusion | option",
     );
     expect(prompt).toContain(
       "known | unknown | provisional | supported | weakened | contradicted | resolved",
     );
     expect(prompt).toContain(
-      "supports | weakens | contradicts | depends_on | causes | may_cause | measures | compares_with | updates | other",
+      "supports | weakens | contradicts | depends_on | causes | may_cause | measures | compares_with | updates | contained_in | other",
     );
   });
 
@@ -635,3 +635,132 @@ describe("buildGraphUpdatePrompt — 57J.59 selected-question contract alignment
     expect(prompt).toContain("genuinely new concepts");
   });
 });
+
+// ── Experiment 60A.3 — Decision Option Prompt Rules ──────────────
+
+describe("60A.3 decision option prompt rules", () => {
+  const prompt = buildGraphUpdatePrompt(makeContext());
+  const section = prompt.split("## Decision Option Structure Rules")[1].split(
+    "## Contract:",
+  )[0];
+
+  it("rule: when user presents mutually exclusive candidate actions, create one option node per alternative", () => {
+    expect(section).toContain("mutually exclusive");
+    expect(section).toContain("option");
+    expect(section).toContain("each");
+  });
+
+  it("rule: keep one unknown as decision context; do not duplicate into every option", () => {
+    expect(section).toContain("unknown");
+    expect(section).toMatch(/decision[-\s]context/);
+  });
+
+  it("rule: link each option to decision-context unknown using contained_in", () => {
+    expect(section).toContain("contained_in");
+    expect(section).toContain("option");
+    expect(section).toContain("unknown");
+  });
+
+  it("rule: attach consequences to the specific option they belong to", () => {
+    expect(section).toContain("specific option");
+    expect(section).toContain("consequence");
+  });
+
+  it("rule text uses double-quoted enum values (no backticks in source)", () => {
+    // The rule section must use "kind" style quoting that works inside a template literal.
+    expect(section).toContain('kind "unknown"');
+    expect(section).toContain('kind "option"');
+    expect(section).toContain('"contained_in"');
+  });
+
+  it("rule explicitly says do not add an alternative_to edge", () => {
+    expect(section).toContain("alternative_to");
+    const lines = section.split("\n").filter((l) => l.includes("alternative_to"));
+    const negLine = lines.find(
+      (l) =>
+        l.toLowerCase().includes("do not") ||
+        l.toLowerCase().includes("not add") ||
+        l.toLowerCase().includes("dont"),
+    );
+    expect(negLine).toBeDefined();
+  });
+
+  it("rule does not require is_baseline / is_default / is_status_quo fields", () => {
+    expect(section).toContain('is_baseline');
+    expect(section).toContain('is_default');
+    expect(section).toContain('is_status_quo');
+    // Verify these appear only in a negation context:
+    const negLines = section.split("\n").filter(
+      (l) => l.includes("is_baseline") || l.includes("is_default"),
+    );
+    expect(negLines.some((l) => l.toLowerCase().includes("do not introduce"))).toBe(true);
+  });
+
+  it("rule explicitly forbids alternative_to edge", () => {
+    const rulesSection = prompt.split("## Decision Option Structure Rules")[1];
+    expect(rulesSection).toContain("alternative_to");
+    // Should say something about not adding it (e.g., "Do not add")
+    const lines = rulesSection.split("\n").filter((l) => l.includes("alternative_to"));
+    const altLine = lines.find((l) =>
+      l.toLowerCase().includes("do not") || l.toLowerCase().includes("not"),
+    );
+    expect(altLine).toBeDefined();
+  });
+
+  it("rule: do-nothing / stay-put is an option when genuinely one of the alternatives", () => {
+    expect(section).toContain("do-nothing") || section.includes("do nothing");
+    // Verify at least one variant is present
+    const hasDoNothing = section.includes("do-nothing") || section.includes("do nothing");
+    expect(hasDoNothing).toBe(true);
+  });
+
+  // Negative checks
+  it("no relocation/savings-specific wording introduced", () => {
+    expect(prompt).not.toContain("relocation");
+    expect(prompt).not.toContain("savings");
+    expect(prompt).not.toContain("engineer");
+    expect(prompt).not.toContain("£2m");
+    expect(prompt).not.toContain("£2M");
+  });
+
+  it("no new decision node kind requirement in prompt", () => {
+    const rulesSection = prompt.split("## Decision Option Structure Rules")[1];
+    // The section must NOT instruct adding a new "decision" node kind — but may mention "decision" as context.
+    expect(rulesSection).toContain('kind "unknown"');
+    expect(rulesSection).toContain('"option"');
+    // Verify the prohibition is against adding a separate decision node:
+    const lines = rulesSection.split("\n").filter((l) => l.includes("decision") && l.toLowerCase().includes("add"));
+    expect(lines.some((l) => l.toLowerCase().includes("do not"))).toBe(true);
+  });
+
+  it("does not require alternative_to — only forbids it", () => {
+    const rulesSection = prompt.split("## Decision Option Structure Rules")[1];
+    // The section must mention alternative_to (to forbid it) but not require it:
+    expect(rulesSection).toContain("alternative_to");
+    // Verify the mention is in a prohibition context, not an encouragement context:
+    const lines = rulesSection.split("\n").filter((l) => l.includes("alternative_to"));
+    expect(lines.some((l) => l.toLowerCase().includes("do not"))).toBe(true);
+  });
+
+  it("baseline flag words appear only in prohibition context", () => {
+    const rulesSection = prompt.split("## Decision Option Structure Rules")[1];
+    // The strings is_baseline, is_default, is_status_quo are mentioned to forbid them:
+    ["is_baseline", "is_default", "is_status_quo"].forEach((flag) => {
+      expect(rulesSection).toContain(flag);
+      // Verify each appears in a negation context:
+      const lines = rulesSection.split("\n").filter((l) => l.includes(flag));
+      expect(lines.some((l) => l.toLowerCase().includes("do not introduce"))).toBe(true);
+    });
+  });
+
+  it("rule: explicitly forbids alternative_to edge", () => {
+    const rulesSection = prompt.split("## Decision Option Structure Rules")[1];
+    expect(rulesSection).toContain("alternative_to");
+    // Should say something about not adding it (e.g., "Do not add")
+    const lines = rulesSection.split("\n").filter((l) => l.includes("alternative_to"));
+    const altLine = lines.find((l) =>
+      l.toLowerCase().includes("do not") || l.toLowerCase().includes("not"),
+    );
+    expect(altLine).toBeDefined();
+  });
+});
diff --git a/tests/graph/schema.test.js b/tests/graph/schema.test.js
index 9603347..5d23e8c 100644
--- a/tests/graph/schema.test.js
+++ b/tests/graph/schema.test.js
@@ -452,6 +452,7 @@ describe("enum values completeness", () => {
       "assumption",
       "unknown",
       "conclusion",
+      "option",
     ];
     const actual = Object.values(SituationKind);
     expect(actual).toEqual(expect.arrayContaining(expected));
@@ -482,6 +483,7 @@ describe("enum values completeness", () => {
       "measures",
       "compares_with",
       "updates",
+      "contained_in",
       "other",
     ];
     const actual = Object.values(SituationRelationship);
@@ -495,3 +497,301 @@ describe("enum values completeness", () => {
     expect(actual).toContain("high");
   });
 });
+
+// ── Experiment 60A.3 — Decision Option Vocabulary Boundary ──────────
+
+describe("60A.3 option node kind", () => {
+  it("accepts 'option' as a valid SituationKind on a node", () => {
+    const result = situationNodeSchema.safeParse({
+      id: "n-opt-a",
+      label: "Option A",
+      description: "Candidate path A",
+      kind: "option",
+      status: "unknown",
+      confidence: "medium",
+      value: null,
+      unit: null,
+      evidenceIds: [],
+      dependsOn: [],
+      affects: [],
+      parentId: null,
+      childIds: [],
+    });
+    expect(result.success).toBe(true);
+  });
+
+  it("existing node kinds are still accepted", () => {
+    for (const kind of Object.values(SituationKind)) {
+      const result = situationNodeSchema.safeParse({
+        id: "n-x",
+        label: "Test",
+        description: "Test",
+        kind,
+        status: "known",
+        confidence: "high",
+        value: null,
+        unit: null,
+        evidenceIds: [],
+        dependsOn: [],
+        affects: [],
+        parentId: null,
+        childIds: [],
+      });
+      expect(result.success).toBe(true);
+    }
+  });
+
+  it("still rejects invalid node kind", () => {
+    const result = situationNodeSchema.safeParse({
+      id: "n-x",
+      label: "Test",
+      description: "Test",
+      kind: "decision",
+      status: "known",
+      confidence: "high",
+      value: null,
+      unit: null,
+      evidenceIds: [],
+      dependsOn: [],
+      affects: [],
+      parentId: null,
+      childIds: [],
+    });
+    expect(result.success).toBe(false);
+  });
+
+  it("rejects 'alternative' as a node kind", () => {
+    const result = situationNodeSchema.safeParse({
+      id: "n-x",
+      label: "Test",
+      description: "Test",
+      kind: "alternative",
+      status: "known",
+      confidence: "high",
+      value: null,
+      unit: null,
+      evidenceIds: [],
+      dependsOn: [],
+      affects: [],
+      parentId: null,
+      childIds: [],
+    });
+    expect(result.success).toBe(false);
+  });
+
+  it("rejects 'baseline' as a node kind", () => {
+    const result = situationNodeSchema.safeParse({
+      id: "n-x",
+      label: "Test",
+      description: "Test",
+      kind: "baseline",
+      status: "known",
+      confidence: "high",
+      value: null,
+      unit: null,
+      evidenceIds: [],
+      dependsOn: [],
+      affects: [],
+      parentId: null,
+      childIds: [],
+    });
+    expect(result.success).toBe(false);
+  });
+
+  it("rejects 'outcome' as a node kind", () => {
+    const result = situationNodeSchema.safeParse({
+      id: "n-x",
+      label: "Test",
+      description: "Test",
+      kind: "outcome",
+      status: "known",
+      confidence: "high",
+      value: null,
+      unit: null,
+      evidenceIds: [],
+      dependsOn: [],
+      affects: [],
+      parentId: null,
+      childIds: [],
+    });
+    expect(result.success).toBe(false);
+  });
+});
+
+describe("60A.3 contained_in relationship", () => {
+  it("accepts 'contained_in' as a valid SituationRelationship on an edge", () => {
+    const result = situationEdgeSchema.safeParse({
+      id: "e-opt-a-to-ctx",
+      fromNodeId: "n-opt-a",
+      toNodeId: "n-decision-ctx",
+      relationship: "contained_in",
+      confidence: "medium",
+      description: "Option A belongs to decision context",
+    });
+    expect(result.success).toBe(true);
+  });
+
+  it("existing relationships are still accepted", () => {
+    for (const rel of Object.values(SituationRelationship)) {
+      const result = situationEdgeSchema.safeParse({
+        id: "e-x",
+        fromNodeId: "n-a",
+        toNodeId: "n-b",
+        relationship: rel,
+        confidence: "high",
+        description: "test",
+      });
+      expect(result.success).toBe(true);
+    }
+  });
+
+  it("rejects invalid relationship type", () => {
+    const result = situationEdgeSchema.safeParse({
+      id: "e-x",
+      fromNodeId: "n-a",
+      toNodeId: "n-b",
+      relationship: "contains_option",
+      confidence: "high",
+      description: "test",
+    });
+    expect(result.success).toBe(false);
+  });
+
+  it("rejects 'alternative_to' as a SituationRelationship", () => {
+    const result = situationEdgeSchema.safeParse({
+      id: "e-x",
+      fromNodeId: "n-a",
+      toNodeId: "n-b",
+      relationship: "alternative_to",
+      confidence: "high",
+      description: "test",
+    });
+    expect(result.success).toBe(false);
+  });
+
+  it("rejects 'option_for' as a SituationRelationship", () => {
+    const result = situationEdgeSchema.safeParse({
+      id: "e-x",
+      fromNodeId: "n-a",
+      toNodeId: "n-b",
+      relationship: "option_for",
+      confidence: "high",
+      description: "test",
+    });
+    expect(result.success).toBe(false);
+  });
+});
+
+describe("60A.3 native two-option graph fixture", () => {
+  it("constructs and validates a complete two-option decision graph", () => {
+    const unknownCtx = makeNode({
+      id: "n-decision-ctx",
+      label: "Which path leaves us better off?",
+      description: "Unresolved decision context for the current choice.",
+      kind: "unknown",
+      status: "unknown",
+      confidence: "medium",
+    });
+
+    const optionA = makeNode({
+      id: "n-option-a",
+      label: "Path A — Act",
+      description: "Candidate action A with its own consequences.",
+      kind: "option",
+      status: "unknown",
+      confidence: "medium",
+    });
+
+    const optionB = makeNode({
+      id: "n-option-b",
+      label: "Path B — Do nothing",
+      description: "Candidate action B (do-nothing / stay-put).",
+      kind: "option",
+      status: "unknown",
+      confidence: "medium",
+    });
+
+    const consequenceA = makeNode({
+      id: "n-cons-a",
+      label: "Benefit of Path A",
+      description: "Consequence that applies to option A.",
+      kind: "metric",
+      status: "known",
+      confidence: "high",
+      value: 100,
+    });
+
+    const consequenceB = makeNode({
+      id: "n-cons-b",
+      label: "Cost of Path B",
+      description: "Consequence that applies to option B.",
+      kind: "observation",
+      status: "known",
+      confidence: "high",
+    });
+
+    const graph = makeGraph({
+      centralStatement: "Test two-option decision.",
+      nodes: [unknownCtx, optionA, optionB, consequenceA, consequenceB],
+      edges: [
+        makeEdge({
+          id: "e-opt-a-contained",
+          fromNodeId: optionA.id,
+          toNodeId: unknownCtx.id,
+          relationship: "contained_in",
+          confidence: "high",
+          description: "Option A belongs to decision context",
+        }),
+        makeEdge({
+          id: "e-opt-b-contained",
+          fromNodeId: optionB.id,
+          toNodeId: unknownCtx.id,
+          relationship: "contained_in",
+          confidence: "high",
+          description: "Option B belongs to decision context",
+        }),
+        makeEdge({
+          id: "e-cons-a-to-opt-a",
+          fromNodeId: consequenceA.id,
+          toNodeId: optionA.id,
+          relationship: "supports",
+          confidence: "high",
+          description: "Consequence supports option A",
+        }),
+        makeEdge({
+          id: "e-cons-b-to-opt-b",
+          fromNodeId: consequenceB.id,
+          toNodeId: optionB.id,
+          relationship: "weakens",
+          confidence: "high",
+          description: "Consequence weakens option B",
+        }),
+      ],
+      activeUnknownNodeId: unknownCtx.id,
+      resolvedNodeIds: [],
+      currentSummary: "2 options under 1 decision context.",
+    });
+
+    expect(graph.nodes.length).toBe(5);
+    expect(graph.edges.length).toBe(4);
+
+    // Both options share the same decision-context unknown
+    const optionEdges = graph.edges.filter((e) => e.relationship === "contained_in");
+    expect(optionEdges.length).toBe(2);
+    expect(optionEdges[0].toNodeId).toBe(optionEdges[1].toNodeId);
+
+    // Each option has its own consequence attached
+    const aConsequences = graph.edges.filter(
+      (e) => e.toNodeId === optionA.id,
+    );
+    const bConsequences = graph.edges.filter(
+      (e) => e.toNodeId === optionB.id,
+    );
+    expect(aConsequences.length).toBeGreaterThan(0);
+    expect(bConsequences.length).toBeGreaterThan(0);
+
+    // No decision node kind exists
+    expect(graph.nodes.some((n) => n.kind === "option")).toBe(true);
+    expect(graph.nodes.some((n) => n.kind === "unknown")).toBe(true);
+  });
+});