Files
buzz/desktop/tests/e2e/agent-access-warning.spec.ts
7fb008f934 fix(desktop): explain open agent access (#2561)
## Why

Hack-day feedback exposed a dangerous mismatch between the UI and the
underlying access model. The `Anyone` respond-to mode appeared as a
neutral dropdown choice, while a Buzz agent may act with the files,
accounts, and tools available on the machine where it runs.

People reasonably read this as sharing a bot in a channel. The current
UI did not explain that it can also share the agent's available access.

## What

- Reframes `respond-to` as **agent access** in user-facing UI.
- Uses plain audience labels: **Only me**, **Anyone**, and **Selected
people**.
- Warns for **both** sharing modes, not just `Anyone` — `Selected
people` also hands host access to someone other than the owner, so only
the audience phrase differs:
> Anyone can use this agent to access your computer, including files,
accounts, and connected tools.

> Selected people can use this agent to access your computer, including
files, accounts, and connected tools.
- Names the machine the agent actually runs on. A provider-backed
(remote) agent reads:
> Anyone can use this agent to access the server it runs on, including
any accounts and tools available there.

The remote wording deliberately omits the owner's files — those aren't
theirs to describe on a host they don't own.
- Places the warning below the selector for `Anyone`, but **after** the
people picker for `Selected people`, so it never sits between the user
and the selection they came to make.
- Removes Nostr, harness, pubkey, and `!shutdown` jargon from the
primary decision copy. Direct pubkey entry remains available as an
advanced path.
- Replaces the green open-access avatar dot with an amber warning marker
and accessible text. Selected access uses a separate blue status.
- Aligns the sidebar action and profile field with the same language.
- Records the shared-field disclosure contract in
`desktop/src/features/agents/AGENTS.md` so future surfaces do not
silently omit it.

## Design decisions

**Persistent inline warning, not a confirmation modal.** The setting
does not autosave; the consequence remains visible beside the selection
until the person chooses **Save access**. This gives the information
before commitment without adding a dismiss-and-confirm ritual that would
repeat in every create/edit surface.

**An unknown run location falls back to the local wording.** It does not
hedge with "computer or server". A remote host requires an installed
`buzz-backend-*` provider, and without one `WhereToRunSection` never
renders — so "server" would name a concept the owner has never been
shown. When it *is* remote, they picked that host from the selector
themselves. Surfaces never synthesize a run location they don't have.

**One resolution site, published through context.** `AgentDialog`
resolves the run location (`runLocationForBackend` from
`ManagedAgent.backend`, `runLocationForRunOn` from the create flow's
`WhereToRunDraft`) and publishes it via `AgentRunLocationContext`. It is
not threaded as a prop through `AgentDefinitionDialog` (1047 lines) or
`AgentInstanceEditDialog` (1228 lines) — neither uses the value, and
both are already over the file-size ceiling. Surfaces outside that tree
(`EditRespondToDialog`) pass the prop directly.

The copy follows the writing system's guidance for high-sensitivity
decisions: lead with the material consequence, use plain actor/action
language, keep helper text adjacent and persistent, and never rely on
color alone.

## Scope

Desktop only. The web and mobile clients do not currently expose this
setting. No protocol, gate, runtime, persistence, or backend behavior
changes.

This does not add team-scoped remote agents. It makes the current
local-or-remote access model honest while that product work remains
separate.

## Validation

- `pnpm exec biome check` and `pnpm exec tsc --noEmit` — clean
- `lib/agentAccessWarning.test.mjs` (8/8) — every mode × run-location
copy variant, both resolvers, unknown-reads-as-local, blank `runOn` is
not a provider
- `ui/respondToFieldContract.test.mjs` (8/8) — plain labels, both
warning positions, source-order guard that the `allowlist` warning
follows the picker, helper-not-inline-copy guard
- `agent-access-warning.spec.ts` (3/3) — native local, provider-backed
remote (asserts the server sentence and *not* "your computer"),
persona-backed edit; includes a bounding-box check that the `Selected
people` warning renders below the picker

---------

Signed-off-by: David Hamilton <daveh@squareup.com>
Signed-off-by: Clay Delk <clay.delk@gmail.com>
Co-authored-by: Clay Delk <clay.delk@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-30 13:32:40 +00:00

206 lines
6.9 KiB
TypeScript

import { expect, test } from "@playwright/test";
import { waitForAnimations } from "../helpers/animations";
import { installMockBridge, TEST_IDENTITIES } from "../helpers/bridge";
const SHOTS = "test-results/agent-access-warning";
async function choosePersonaAccess(
page: import("@playwright/test").Page,
optionName: string,
) {
await page.locator("#agent-respond-to").click();
await page.getByRole("menuitemradio", { name: optionName }).click();
}
async function openAgentAccessDialog(
page: import("@playwright/test").Page,
agentPubkey: string,
) {
if (!(await page.getByTestId("members-sidebar").isVisible())) {
await page.getByTestId("channel-general").click();
await page.getByTestId("channel-members-trigger").click();
await expect(page.getByTestId("members-sidebar")).toBeVisible();
}
const row = page.getByTestId(`sidebar-member-${agentPubkey}`);
const menu = page.getByTestId(`sidebar-member-menu-${agentPubkey}`);
await row.hover();
await menu.focus();
await menu.press("Enter");
await page.getByTestId(`sidebar-edit-respond-to-${agentPubkey}`).click();
await expect(
page.getByRole("dialog", { name: "Manage agent access" }),
).toBeVisible();
}
test("open agent access explains the available access before save", async ({
page,
}) => {
const agent = TEST_IDENTITIES.charlie;
await installMockBridge(page, {
managedAgents: [
{
pubkey: agent.pubkey,
name: "Hack Day Helper",
status: "running",
channelNames: ["general"],
respondTo: "owner-only",
},
],
});
await page.goto("/");
await openAgentAccessDialog(page, agent.pubkey);
const accessSelect = page.getByTestId("agent-respond-to-select");
await expect(accessSelect).toHaveValue("owner-only");
await expect(page.getByTestId("agent-access-warning")).toHaveCount(0);
const saveAccess = page.getByRole("button", { name: "Save access" });
await expect(saveAccess).toBeVisible();
const commandsBeforeSave = await page.evaluate(
() => window.__BUZZ_E2E_COMMAND_LOG__?.length ?? 0,
);
await accessSelect.selectOption("anyone");
const warning = page.getByTestId("agent-access-warning");
await expect(warning).toBeVisible();
await expect(warning).toContainText(
"Anyone can use this agent to access your computer, including files, accounts, and connected tools.",
);
await waitForAnimations(page);
await page
.getByRole("dialog", { name: "Manage agent access" })
.screenshot({ path: `${SHOTS}/open-access-warning.png` });
await saveAccess.click();
await expect(
page.getByRole("dialog", { name: "Manage agent access" }),
).not.toBeVisible();
await expect
.poll(async () =>
page.evaluate((start) => {
const commands = window.__BUZZ_E2E_COMMAND_LOG__ ?? [];
return commands
.slice(start)
.some(
(entry) =>
entry.command === "update_managed_agent" &&
(entry.payload as { input?: { respondTo?: string } })?.input
?.respondTo === "anyone",
);
}, commandsBeforeSave),
)
.toBe(true);
await openAgentAccessDialog(page, agent.pubkey);
await expect(accessSelect).toHaveValue("anyone");
// Selected people narrows the audience but not the access, so the warning
// persists with its own audience phrase.
await accessSelect.selectOption("allowlist");
await expect(warning).toBeVisible();
await expect(warning).toContainText(
"Selected people can use this agent to access your computer, including files, accounts, and connected tools.",
);
const picker = page.getByTestId("agent-respond-to-allowlist");
await expect(
picker.getByText("Selected people", { exact: true }),
).toBeVisible();
// The warning sits below the picker so it never blocks the selection the
// user came here to make.
await waitForAnimations(page);
const pickerBox = await picker.boundingBox();
const warningBox = await warning.boundingBox();
expect(pickerBox?.y).toBeDefined();
expect(warningBox?.y).toBeGreaterThan(pickerBox?.y ?? 0);
await page
.getByRole("dialog", { name: "Manage agent access" })
.screenshot({ path: `${SHOTS}/selected-people-warning.png` });
// Only me shares nothing, so the warning goes away entirely.
await accessSelect.selectOption("owner-only");
await expect(warning).toHaveCount(0);
});
test("a provider-backed agent's warning names the server, not this computer", async ({
page,
}) => {
const agent = TEST_IDENTITIES.charlie;
await installMockBridge(page, {
managedAgents: [
{
pubkey: agent.pubkey,
name: "Remote Helper",
status: "running",
channelNames: ["general"],
respondTo: "owner-only",
backend: { type: "provider", id: "blox", config: {} },
},
],
});
await page.goto("/");
await openAgentAccessDialog(page, agent.pubkey);
await page.getByTestId("agent-respond-to-select").selectOption("anyone");
const warning = page.getByTestId("agent-access-warning");
await expect(warning).toContainText(
"Anyone can use this agent to access the server it runs on, including any accounts and tools available there.",
);
// The local wording must not leak into a remote-backed agent.
await expect(warning).not.toContainText("your computer");
});
test("persona-backed edit warns before saving open access", async ({
page,
}) => {
const agent = TEST_IDENTITIES.tyler;
await installMockBridge(page, {
managedAgents: [
{
pubkey: agent.pubkey,
name: "Tyler Agent",
status: "stopped",
channelNames: ["agents"],
respondTo: "owner-only",
},
],
});
await page.goto("/");
await page.getByTestId("open-agents-view").click();
await page.getByRole("button", { name: "Tyler Agent agent profile" }).click();
await page.getByTestId("user-profile-edit-agent").click();
const dialog = page.getByTestId("edit-agent-dialog");
await expect(dialog).toBeVisible();
await expect(page.locator("#agent-respond-to")).toHaveText(
"Only me (default)",
);
await choosePersonaAccess(page, "Anyone");
await expect(dialog.getByTestId("agent-access-warning")).toContainText(
"Anyone can use this agent to access your computer, including files, accounts, and connected tools.",
);
const commandsBeforeSave = await page.evaluate(
() => window.__BUZZ_E2E_COMMAND_LOG__?.length ?? 0,
);
await page.getByTestId("edit-agent-dialog-submit").click();
await expect(dialog).not.toBeVisible();
await expect
.poll(async () =>
page.evaluate((start) => {
const commands = window.__BUZZ_E2E_COMMAND_LOG__ ?? [];
return commands
.slice(start)
.some(
(entry) =>
entry.command === "update_managed_agent" &&
(entry.payload as { input?: { respondTo?: string } })?.input
?.respondTo === "anyone",
);
}, commandsBeforeSave),
)
.toBe(true);
});