Compare commits

...
Author SHA1 Message Date
ba765ba3f2 test(server): reproduce stale plugin history after rewind
Preserve the manager-level reproduction for #4667 and a file-only
control. A second red case shows that discarded timeline snapshots
also shorten replacement output when an item identity is reused.
These are diagnostic tests; production behavior is unchanged.

Refs #4667

Co-authored-by: Matt Cowger <matt@cowger.us>
Co-authored-by: Omer Cohen <639682+omercnet@users.noreply.github.com>
2026-10-27 18:22:30 +00:00
@@ -229,6 +229,149 @@ function eventsOfType(events: AgentStreamEvent[], type: AgentStreamEvent["type"]
return events.filter((event) => event.type === type);
}
function createRewindProviderHarness(options: { reuseAnswerId?: boolean } = {}) {
const nativeMessages: string[] = [];
const harness = createProviderHarness({
capabilities: [...CAPABILITIES, "session.revert.conversation", "session.revert.files"],
async handleInput(input, emit) {
if (input.type === "session.prompt") {
if (input.prompt.input.type !== "message") throw new Error("Expected a message");
const text = input.prompt.input.content
.flatMap((part) => (part.type === "text" ? [part.text] : []))
.join("");
nativeMessages.push(text);
const turnId = `turn-${text}`;
emit({
type: "session.prompt_result",
sessionId: input.sessionId,
clientMessageId: input.prompt.clientMessageId,
result: { type: "turn", turnId },
});
emit({ type: "session.turn", sessionId: input.sessionId, turnId, state: "started" });
emit({
type: "timeline.item",
sessionId: input.sessionId,
item: {
type: "user_message",
id: `user-${text}`,
text,
clientMessageId: input.prompt.clientMessageId,
revertToken: text,
},
});
emit({
type: "timeline.item",
sessionId: input.sessionId,
item: {
type: "assistant_message",
id: options.reuseAnswerId ? `answer-${nativeMessages.length}` : `answer-${text}`,
text: `Reply ${text}`,
},
});
emit({ type: "session.turn", sessionId: input.sessionId, turnId, state: "completed" });
return true;
}
if (input.type === "session.revert") {
if (input.scope !== "files")
nativeMessages.splice(nativeMessages.indexOf(String(input.token)));
emit({ type: "request.completed", requestId: input.requestId });
return true;
}
return false;
},
});
return { ...harness, nativeMessages };
}
async function withRewindAgent(
run: (context: {
manager: AgentManager;
agentId: string;
nativeMessages: string[];
messages: () => string[];
}) => Promise<void>,
) {
const harness = createRewindProviderHarness();
const logger = createTestLogger();
const registry = new PluginAgentClientRegistry(logger);
registry.replace([harness.registration]);
const manager = new AgentManager({
logger,
clients: registry.clients(),
providerDefinitions: registry.definitions(),
});
const agent = await manager.createAgent(
{ provider: harness.registration.id, cwd: tmpdir() },
undefined,
{ workspaceId: undefined },
);
const messages = () =>
manager
.getTimeline(agent.id)
.flatMap((item) =>
item.type === "user_message" || item.type === "assistant_message" ? [item.text] : [],
);
try {
await manager.runAgent(agent.id, "A", { clientMessageId: "client-A" });
await manager.runAgent(agent.id, "B", { clientMessageId: "client-B" });
await manager.runAgent(agent.id, "C", { clientMessageId: "client-C" });
expect(messages()).toEqual(["A", "Reply A", "B", "Reply B", "C", "Reply C"]);
await run({ manager, agentId: agent.id, nativeMessages: harness.nativeMessages, messages });
} finally {
await manager.closeAgent(agent.id);
await registry.shutdown();
}
}
test("conversation rewind discards withdrawn plugin turns before and after a replacement prompt", async () => {
await withRewindAgent(async ({ manager, agentId, nativeMessages, messages }) => {
await manager.rewind(agentId, "user-B", "conversation");
const afterRewind = messages();
expect(nativeMessages).toEqual(["A"]);
await manager.runAgent(agentId, "D", { clientMessageId: "client-D" });
const afterReplacement = messages();
expect(nativeMessages).toEqual(["A", "D"]);
expect({ afterRewind, afterReplacement }).toEqual({
afterRewind: ["A", "Reply A"],
afterReplacement: ["A", "Reply A", "D", "Reply D"],
});
});
});
test("file-only plugin rewind preserves the conversation before and after a replacement prompt", async () => {
await withRewindAgent(async ({ manager, agentId, nativeMessages, messages }) => {
await manager.rewind(agentId, "user-B", "files");
expect(nativeMessages).toEqual(["A", "B", "C"]);
expect(messages()).toEqual(["A", "Reply A", "B", "Reply B", "C", "Reply C"]);
await manager.runAgent(agentId, "D", { clientMessageId: "client-D" });
expect(messages()).toEqual(["A", "Reply A", "B", "Reply B", "C", "Reply C", "D", "Reply D"]);
});
});
test("plugin rewind restores translation state when a replacement reuses a discarded item id", async () => {
const harness = createRewindProviderHarness({ reuseAnswerId: true });
const registry = new PluginAgentClientRegistry(createTestLogger());
registry.replace([harness.registration]);
const client = registry.clients()[harness.registration.id]!;
const session = await client.createSession({ provider: harness.registration.id, cwd: tmpdir() });
const live: AgentStreamEvent[] = [];
try {
await session.run("A");
await session.run("B");
await session.revertConversation!({ messageId: "user-B" });
session.subscribe((event) => live.push(event));
await session.run("B revised");
const assistant = live.flatMap((event) =>
event.type === "timeline" && event.item.type === "assistant_message" ? [event.item.text] : [],
);
expect(harness.nativeMessages).toEqual(["A", "B revised"]);
expect(assistant).toEqual(["Reply B revised"]);
} finally {
await session.close();
await registry.shutdown();
}
});
function openNestedChildren(rootSessionId: string, emit: (event: ProviderEvent) => void) {
for (const [sessionId, parentSessionId] of [
["a", rootSessionId],