Files
paseo/packages/app/e2e/browser/permission-steer.real.spec.ts
T
c6b04f302a fix(server): stop a steered prompt from deadlocking on a permission (#3585)
* fix(server): stop a steered prompt from deadlocking on a permission

Claude parks its turn inside the blocking canUseTool callback while a
plan approval or a tool prompt waits for an answer. The SDK stops to
read its input stream there, so a message sent at the open card is
queued and never read. The user must reject the card or interrupt the
agent by hand, and then send the message a second time.

Deny the pending permissions after the steer is admitted. The prompt is
already in the SDK input queue when the permission callback returns, so
Claude reads it in the same turn. Plan mode stays on and no turn
restarts.

One pass is not enough. A turn that made several tool calls opens its
next request 10 to 300 milliseconds after the previous one is answered,
so a single snapshot advances the batch by one and parks the turn again.
The Claude adapter already tracks the steers the SDK has not read, so
expose that as an optional AgentSession.hasUnreadSteer and answer each
new request while that method reports the prompt as unread. The release
stops the moment the prompt is read, so a permission the agent asks for
after it acts on the message still reaches the user.

Only the two entrypoints that carry a message from a person set the
flag. A scheduled prompt, a child-finished notification, and an
agent-to-agent prompt must not answer a card the user is looking at.

Denial left no timeline row, so a denied plan disappeared from the
transcript. Claude and Codex now push a plan_approval row with
approved: false. For Codex the row comes from resolvePlanPermission,
which is the path that dismissPendingPlanApprovals also uses.

In the app, queue mode held the message until the turn ended, and a
parked turn never ends. The composer now sends while a permission is
pending, and the button label agrees with the submit path.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(server): keep a post-prompt permission out of the release

The release loop answered the pending requests before it asked whether
the agent had read the prompt. The agent can read the prompt during a
poll interval and then open a permission request because of it. The next
pass denied that request, so a decision that belongs to the user was
made for them.

The loop now asks first and answers second. The first pass stays
unconditional, because a provider that does not report unread steers
must still get one release.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(server): make steer permission clearing provider-owned

Keep the daemon policy explicit while letting each provider queue the steer and release permissions that block it. This removes timing-based polling and preserves rejected plans as canonical timeline rows.

* fix(server): preserve omitted steer options

Only construct provider steer options when permission clearing is requested. System prompts otherwise keep the existing undefined options contract.

---------

Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: Mohamed Boudra <boudra.moha@gmail.com>
2026-08-20 14:02:00 +02:00

84 lines
3.3 KiB
TypeScript

import { mkdtempSync, realpathSync } from "node:fs";
import { tmpdir } from "node:os";
import path from "node:path";
import { expect, test } from "../support/fixtures";
import { submitMessage } from "../support/helpers/composer";
import { waitForPermissionPrompt } from "../support/helpers/permissions";
import { cleanupRewindFlow, type AgentHandle, launchAgent } from "../support/helpers/rewind-flow";
const scenarios = [
{
provider: "claude" as const,
providerConfig: { model: "haiku", modeId: "plan" },
planPrompt:
"Do not ask questions. Produce a concise plan to append the line 'Hello' to README.md, then present it for approval. Do not implement it.",
planText: "README.md",
steerPrompt: "Do not implement anything. Reply exactly CLAUDE_STEER_RECEIVED and stop.",
reply: "CLAUDE_STEER_RECEIVED",
},
{
provider: "codex" as const,
providerConfig: { featureValues: { plan_mode: true } },
planPrompt:
"Produce a concise implementation plan with exactly these steps: Inspect permission steering; Implement permission steering; Verify permission steering. Do not implement it.",
planText: "Inspect permission steering",
steerPrompt: "Do not produce another plan. Reply exactly CODEX_STEER_RECEIVED and stop.",
reply: "CODEX_STEER_RECEIVED",
},
];
test.describe("composer steer supersedes plan approval", () => {
for (const scenario of scenarios) {
test(`${scenario.provider} keeps the rejected plan in the timeline`, async ({
page,
}, testInfo) => {
test.setTimeout(420_000);
const cwd = realpathSync(
mkdtempSync(path.join(tmpdir(), `paseo-permission-steer-${scenario.provider}-`)),
);
let handle: AgentHandle | undefined;
try {
handle = await launchAgent({
page,
provider: scenario.provider,
cwd,
mode: "full-access",
providerConfig: scenario.providerConfig,
});
await submitMessage(page, scenario.planPrompt);
await waitForPermissionPrompt(page, 180_000);
const pendingPlan = page.getByTestId("permission-plan-card");
await expect(pendingPlan).toContainText(scenario.planText);
const pendingScreenshot = testInfo.outputPath(`${scenario.provider}-pending-plan.png`);
await page.screenshot({ path: pendingScreenshot });
await testInfo.attach(`${scenario.provider} pending plan`, {
path: pendingScreenshot,
contentType: "image/png",
});
await submitMessage(page, scenario.steerPrompt);
await expect(pendingPlan).toHaveCount(0, { timeout: 30_000 });
const rejectedPlan = page.getByTestId("timeline-plan-card");
await expect(rejectedPlan).toBeVisible({ timeout: 30_000 });
await expect(rejectedPlan).toContainText(scenario.planText);
await expect(
page.getByTestId("assistant-message").filter({ hasText: scenario.reply }),
).toBeVisible({
timeout: 180_000,
});
const rejectedScreenshot = testInfo.outputPath(`${scenario.provider}-rejected-plan.png`);
await page.screenshot({ path: rejectedScreenshot });
await testInfo.attach(`${scenario.provider} rejected plan`, {
path: rejectedScreenshot,
contentType: "image/png",
});
} finally {
await cleanupRewindFlow({ handle, cwd });
}
});
}
});