Skip to content

Stabilize cloud E2E scenarios #1735

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
RhysSullivan merged 1 commit into from
Aug 25, 2026
Merged

Stabilize cloud E2E scenarios #1735

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion .github/workflows/ci.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -304,7 +304,9 @@ jobs:
if: matrix.target == 'cloud'
env:
MCP_SESSION_TIMEOUT_MS: "3000"
MCP_PAUSED_SESSION_IDLE_TIMEOUT_MS: "6000"
# Still 18x shorter than production, but long enough for a cold Vite
# resume route to compile under a fully loaded CI runner.
MCP_PAUSED_SESSION_IDLE_TIMEOUT_MS: "30000"
run: bun scripts/run-ci-shard.ts cloud ${{ matrix['shard-index'] }}
working-directory: e2e

Expand Down
122 changes: 64 additions & 58 deletions e2e/cloud/mcp-browser-resume-page.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,6 @@ import { scenario } from "../src/scenario";
import { Api, Browser, Mcp, Target } from "../src/services";
import { parseBrowserApproval } from "../src/surfaces/mcp";
import type { Identity } from "../src/target";
import { visit } from "../src/surfaces/browser";

const coreApi = composePluginApi([] as const);

Expand Down Expand Up @@ -196,68 +195,75 @@ scenario(
openBrowserApprovalSession(target.mcpUrl, bearer),
);
yield* Effect.gen(function* () {
const paused = yield* Effect.promise(() =>
session.client.callTool({ name: "execute", arguments: { code: GATED_CODE } }),
);
const approval = parseBrowserApproval({
raw: paused,
text: textOf(paused),
ok: paused.isError !== true,
});
yield* browser.session(identity, async ({ page, step }) => {
// Open Chromium before pausing. The suite intentionally compresses
// the production nine-minute decision window to 30 seconds, and a
// cold browser launch should not consume the user's decision time.
const paused = await session.client.callTool({
name: "execute",
arguments: { code: GATED_CODE },
});
const approval = parseBrowserApproval({
raw: paused,
text: textOf(paused),
ok: paused.isError !== true,
});

const approvalUrl = new URL(approval.approvalUrl);
const mcpSessionId = approvalUrl.searchParams.get("mcp_session_id");
expect(
mcpSessionId,
"approval URL carries the MCP session id that the browser page will query",
).toEqual(expect.any(String));
expect(mcpSessionId, "approval URL points at the session that paused").toBe(
session.transport.sessionId,
);

const [resumed] = yield* Effect.all(
[
Effect.promise(() =>
session.client.callTool({
name: "resume",
arguments: { executionId: approval.executionId },
}),
),
browser.session(identity, async ({ page, step }) => {
await step("Open the paused execution approval page", async () => {
await visit(page, pathWithSearch(approval.approvalUrl));
await page.getByText("User approval required").waitFor();
});
const approvalUrl = new URL(approval.approvalUrl);
const mcpSessionId = approvalUrl.searchParams.get("mcp_session_id");
expect(
mcpSessionId,
"approval URL carries the MCP session id that the browser page will query",
).toEqual(expect.any(String));
expect(mcpSessionId, "approval URL points at the session that paused").toBe(
session.transport.sessionId,
);

await step("Review the paused tool call details", async () => {
await page.getByText("Pending request").waitFor();
await page.getByText(/Approve executor\.coreTools\.policies\.list\?/).waitFor();
await step("Approve the paused tool call through the browser page", async () => {
// The resume GET is this page's concrete readiness signal. The
// generic browser `visit` helper additionally waits up to five
// seconds for network-idle, which is inappropriate while this
// deliberately short approval lease is ticking.
const pausedExecutionLoaded = page.waitForResponse((response) => {
const responseUrl = new URL(response.url());
return (
response.request().method() === "GET" &&
responseUrl.pathname.endsWith(
`/api/mcp-sessions/${mcpSessionId}/executions/${approval.executionId}`,
) &&
response.status() === 200
);
});
await page.goto(pathWithSearch(approval.approvalUrl), { waitUntil: "load" });
await pausedExecutionLoaded;
await page.getByText("User approval required").waitFor();
await page.getByText("Pending request").waitFor();
await page.getByText(/Approve executor\.coreTools\.policies\.list\?/).waitFor();

const approve = page.getByRole("button", { name: "Approve" });
await approve.waitFor();
expect(
await approve.isEnabled(),
"the approve control is enabled for the paused execution",
).toBe(true);
expect(
await page.getByText(UNAVAILABLE_COPY).count(),
"the resume page does not show the expired-session failure copy",
).toBe(0);
});
const approve = page.getByRole("button", { name: "Approve" });
await approve.waitFor();
expect(
await approve.isEnabled(),
"the approve control is enabled for the paused execution",
).toBe(true);
expect(
await page.getByText(UNAVAILABLE_COPY).count(),
"the resume page does not show the expired-session failure copy",
).toBe(0);
await page.getByRole("button", { name: "Approve" }).click();
await page.getByText("Approve sent").waitFor();
});

await step("Approve the paused tool call", async () => {
await page.getByRole("button", { name: "Approve" }).click();
await page.getByText("Approve sent").waitFor();
});
}),
],
{ concurrency: "unbounded" },
);
const resumed = await session.client.callTool({
name: "resume",
arguments: { executionId: approval.executionId },
});

expect(resumed.isError, "browser-mode resume completed after the UI approval").not.toBe(
true,
);
expect(textOf(resumed), "the gated tool completed after approval").toContain(policy.id);
expect(resumed.isError, "browser-mode resume completed after the UI approval").not.toBe(
true,
);
expect(textOf(resumed), "the gated tool completed after approval").toContain(policy.id);
});
}).pipe(Effect.ensuring(closeQuietly(session)));
}).pipe(
Effect.ensuring(
Expand Down
Loading
Loading