robbond/confidence-engine
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

test(reasoning): add customer signing followup fixture

Browse Source
This commit is contained in:
robbond
2026-08-14 08:02:45 +01:00
parent 870d325d08
commit bcbcb65020
2 changed files with 195 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/fixtures/pre-anchored-product-launch-customer-signing.json
Vendored
+126
View File
@@ -0,0 +1,126 @@
{
"description": "Deterministic pre-anchored product-launch customer-signing follow-up fixture — represents the confirmed state immediately before the material customer-signing answer.",
"scenario": "We are evaluating two product-launch timing options: launching the new software product this year or waiting twelve months.",
"unresolved_question": "Which option leaves us better off overall?",
"selectedQuestion": {
"nodeId": "n_enterprise_customer_signing",
"question": "What evidence would clarify whether one prospective enterprise customer will sign if we launch this year?",
"reason": "decision"
},
"graph": {
"centralStatement": "We are evaluating two product-launch timing options: launching the new software product this year or waiting twelve months.",
"nodes": [
{
"id": "n_product_launch_state",
"label": "Product launch timing consideration",
"description": "Current state: the organisation is evaluating whether to launch the new software product this year or wait twelve months.",
"kind": "state",
"status": "provisional",
"confidence": "high",
"value": null,
"unit": null,
"evidenceIds": [],
"dependsOn": [],
"affects": [],
"parentId": null,
"childIds": []
},
{
"id": "opt_launch_this_year",
"label": "Launch this year",
"description": "Launch the new software product this year. Consequences: £1.2M/year expected additional recurring revenue, £300k one-off launch/support cost.",
"kind": "option",
"status": "known",
"confidence": "high",
"value": null,
"unit": null,
"evidenceIds": [],
"dependsOn": [],
"affects": [],
"parentId": null,
"childIds": []
},
{
"id": "opt_wait_twelve_months",
"label": "Wait twelve months",
"description": "Wait twelve months before launching the product. Consequences: avoids the £300k immediate launch/support cost, forgoes approximately £1.2M of first-year revenue.",
"kind": "option",
"status": "known",
"confidence": "high",
"value": null,
"unit": null,
"evidenceIds": [],
"dependsOn": [],
"affects": [],
"parentId": null,
"childIds": []
},
{
"id": "n_product_launch_decision",
"label": "Which option leaves us better off overall?",
"description": "Uncertainty about which of the two product-launch timing options — launch this year or wait twelve months — provides superior net value for the organisation.",
"kind": "unknown",
"status": "unknown",
"confidence": "medium",
"value": null,
"unit": null,
"evidenceIds": [],
"dependsOn": [],
"affects": [],
"parentId": null,
"childIds": []
},
{
"id": "n_enterprise_customer_signing",
"label": "Prospective enterprise customer signing status",
"description": "Unknown whether one prospective enterprise customer will sign if we launch this year, because they account for approximately £700,000 of the £1.2 million expected annual revenue.",
"kind": "unknown",
"status": "unknown",
"confidence": "medium",
"value": null,
"unit": null,
"evidenceIds": [],
"dependsOn": [],
"affects": [],
"parentId": null,
"childIds": []
}
],
"edges": [
{
"id": "e-opt-launch-to-dec",
"fromNodeId": "opt_launch_this_year",
"toNodeId": "n_product_launch_decision",
"relationship": "contained_in",
"confidence": "high",
"description": "Launch this year option is a candidate for the product launch decision"
},
{
"id": "e-opt-wait-to-dec",
"fromNodeId": "opt_wait_twelve_months",
"toNodeId": "n_product_launch_decision",
"relationship": "contained_in",
"confidence": "high",
"description": "Wait twelve months option is a candidate for the product launch decision"
},
{
"id": "e-customer-signing-to-launch-option",
"fromNodeId": "n_enterprise_customer_signing",
"toNodeId": "opt_launch_this_year",
"relationship": "contained_in",
"confidence": "high",
"description": "Prospective enterprise customer signing status is material to the launch this year option"
}
],
"activeUnknownNodeId": "n_enterprise_customer_signing",
"resolvedNodeIds": [],
"currentSummary": "Two product-launch timing options evaluated; net-value comparison unresolved; prospective enterprise customer signing remains the active material uncertainty.",
"reasoningState": {
"comparabilityStatus": null,
"relationshipStatus": null,
"relationshipAssessed": false,
"contradictionReasoningAllowed": true,
"reasoningStages": []
}
}
}
tests/reproduce-multi-turn-investigation.harness.test.js
+69
View File
@@ -2,6 +2,11 @@ import { describe, it, expect } from "vitest";
import fs from "fs";
import { fileURLToPath } from "url";
import path from "path";
import { situationGraphSchema } from "@/lib/graph/schema.js";
import {
validateGraphReferences,
detectDuplicateNodeIds,
} from "@/lib/graph/utils.js";
// ── Load committed decision-options fixture directly (60A.7) ─────────────
const DECISION_OPTIONS_FIXTURE_PATH = path.resolve(
@@ -10,6 +15,14 @@ const DECISION_OPTIONS_FIXTURE_PATH = path.resolve(
);
const DECISION_OPTIONS_FIXTURE = JSON.parse(fs.readFileSync(DECISION_OPTIONS_FIXTURE_PATH, "utf-8"));
const PRODUCT_LAUNCH_CUSTOMER_SIGNING_FIXTURE_PATH = path.resolve(
path.dirname(fileURLToPath(import.meta.url)),
"fixtures/pre-anchored-product-launch-customer-signing.json",
);
const PRODUCT_LAUNCH_CUSTOMER_SIGNING_FIXTURE = JSON.parse(
fs.readFileSync(PRODUCT_LAUNCH_CUSTOMER_SIGNING_FIXTURE_PATH, "utf-8"),
);
/**
* Synchronous simulator of the pre-anchored update-only mode with a custom fixture.
* Mirrors what reproduce-multi-turn-investigation.mjs does when FIXTURE_MODE is set
@@ -1372,6 +1385,62 @@ describe("reproduce-multi-turn-investigation harness: one-shot semantics", () =>
});
});
describe("pre-anchored product-launch customer-signing fixture", () => {
it("parses, validates, preserves identities, and exposes the customer unknown as the active target", () => {
const fixture = PRODUCT_LAUNCH_CUSTOMER_SIGNING_FIXTURE;
const graph = situationGraphSchema.parse(fixture.graph);
const graphReferenceValidation = validateGraphReferences(graph);
const duplicateNodeIds = detectDuplicateNodeIds(graph.nodes);
const decisionNodes = graph.nodes.filter((node) => node.id === "n_product_launch_decision");
const launchOptionNodes = graph.nodes.filter((node) => node.id === "opt_launch_this_year");
const waitOptionNodes = graph.nodes.filter((node) => node.id === "opt_wait_twelve_months");
const customerUnknownNodes = graph.nodes.filter(
(node) => node.id === "n_enterprise_customer_signing",
);
const additionalUnknownIds = graph.nodes
.filter(
(node) =>
node.kind === "unknown" &&
![
"n_product_launch_decision",
"n_enterprise_customer_signing",
].includes(node.id),
)
.map((node) => node.id);
expect(fixture.selectedQuestion?.question).toBe(
"What evidence would clarify whether one prospective enterprise customer will sign if we launch this year?",
);
expect(fixture.selectedQuestion?.nodeId).toBe("n_enterprise_customer_signing");
expect(decisionNodes).toHaveLength(1);
expect(decisionNodes[0].status).toBe("unknown");
expect(launchOptionNodes).toHaveLength(1);
expect(waitOptionNodes).toHaveLength(1);
expect(customerUnknownNodes).toHaveLength(1);
expect(customerUnknownNodes[0].status).toBe("unknown");
expect(customerUnknownNodes[0].kind).toBe("unknown");
expect(additionalUnknownIds).toEqual([]);
expect(graphReferenceValidation.valid).toBe(true);
expect(duplicateNodeIds).toEqual([]);
expect(graph.activeUnknownNodeId).toBe("n_enterprise_customer_signing");
expect(customerUnknownNodes[0].id).toBe(graph.activeUnknownNodeId);
const customerLinkage = graph.edges.filter(
(edge) =>
edge.fromNodeId === "n_enterprise_customer_signing" &&
edge.toNodeId === "opt_launch_this_year" &&
edge.relationship === "contained_in",
);
expect(customerLinkage).toHaveLength(1);
});
});
// ── Pre-anchored update-only mode (57J.74) ─────────────
/**