Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
import { describe, it, expect, vi } from 'vitest';
const mocks = vi.hoisted(() => ({
getVoiceConfig: vi.fn(async () => ({
enabled: true,
llm: {
model: 'test-model',
provider: 'test-provider',
usePersonality: false,
systemPrompt: 'sys',
tools: { enabled: false, maxIterations: 1 },
},
})),
synthesize: vi.fn(async (text) => ({ wav: Buffer.alloc(8), latencyMs: 1, _text: text })),
streamChat: vi.fn(),
}));
vi.mock('./config.js', () => ({ getVoiceConfig: mocks.getVoiceConfig }));
vi.mock('./stt.js', () => ({ transcribe: vi.fn() }));
vi.mock('./tts.js', () => ({ synthesize: (...args) => mocks.synthesize(...args) }));
vi.mock('./llm.js', () => ({ streamChat: (...args) => mocks.streamChat(...args) }));
vi.mock('./tools.js', () => ({
getToolSpecsForIntent: () => ({ specs: undefined, activeGroups: new Set() }),
classifyIntent: () => new Set(),
dispatchTool: vi.fn(),
getAllToolNames: () => [],
UI_KINDS: ['tab', 'button', 'link', 'input', 'textarea', 'select', 'checkbox', 'radio'],
}));
vi.mock('./echo.js', () => ({
isEchoOfRecentTts: () => false,
rememberTtsSentence: vi.fn(),
}));
vi.mock('../brainJournal.js', () => ({ appendJournal: vi.fn(), getToday: vi.fn() }));
vi.mock('./confirmGate.js', () => ({ resolvePending: vi.fn(), isExpired: vi.fn() }));
const { runTurn } = await import('./pipeline.js');
describe('runTurn LLM timeout recovery', () => {
it('cancels queued TTS without converting the timeout into user cancellation', async () => {
mocks.synthesize.mockClear();
mocks.streamChat.mockImplementationOnce(async (_messages, { onDelta, onTimeout }) => {
onDelta('The provider stalled.');
onTimeout();
throw new Error('Voice LLM request timed out');
});
const events = [];
const signal = new AbortController().signal;
const turn = runTurn({
text: 'hello',
history: [],
emit: (event, payload) => events.push({ event, payload }),
signal,
state: {},
});
await expect(turn).rejects.toThrow('Voice LLM request timed out');
await vi.waitFor(() => expect(mocks.synthesize).not.toHaveBeenCalled());
expect(signal.aborted).toBe(false);
expect(events).toContainEqual({
event: 'voice:tts:cancel',
payload: { reason: 'timeout' },
});
expect(events.some(({ event }) => event === 'voice:tts:audio')).toBe(false);
});
});