@@ -1,50 +1,50 @@
11import { describe, expect, it } from "vitest";
2-import { hasModelSwitchContinuityEvidence } from "./model-switch-eval.js";
2+import { hasModelSwitchContinuitySignal } from "./model-switch-eval.js";
3344describe("qa model-switch evaluation", () => {
55it("accepts direct handoff replies that mention the kickoff task", () => {
66expect(
7-hasModelSwitchContinuityEvidence(
7+hasModelSwitchContinuitySignal(
88"Handoff confirmed: I reread QA_KICKOFF_TASK.md and switched to gpt.",
99),
1010).toBe(true);
1111});
12121313it("accepts short mission-oriented switch confirmations", () => {
1414expect(
15-hasModelSwitchContinuityEvidence(
15+hasModelSwitchContinuitySignal(
1616"model switch complete. reread the kickoff task; qa mission stays the same.",
1717),
1818).toBe(true);
1919});
20202121it("accepts concise kickoff note confirmations", () => {
2222expect(
23-hasModelSwitchContinuityEvidence(
23+hasModelSwitchContinuitySignal(
2424"Handoff clean: after the model switch, I reread the kickoff note.",
2525),
2626).toBe(true);
2727});
28282929it("accepts concise paraphrases of the kickoff task after a handoff", () => {
3030expect(
31-hasModelSwitchContinuityEvidence(
31+hasModelSwitchContinuitySignal(
3232"Handoff is clear: after the model switch, read source and docs first, run seeded qa-channel scenarios, and report worked, failed, blocked, and follow-up.",
3333),
3434).toBe(true);
3535});
36363737it("rejects unrelated handoff chatter that never confirms the kickoff reread", () => {
3838expect(
39-hasModelSwitchContinuityEvidence(
39+hasModelSwitchContinuitySignal(
4040"subagent-handoff confirmed. qa report update: scenario pass. qa run complete.",
4141),
4242).toBe(false);
4343});
44444545it("rejects over-scoped multi-line wrap-ups even if they mention a switch and the mission", () => {
4646expect(
47-hasModelSwitchContinuityEvidence(
47+hasModelSwitchContinuitySignal(
4848`model switch acknowledged. qa mission stays the same.
49495050Final QA tally update: all mandatory scenarios resolved. QA run complete.`,
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。