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
/**
* Pipeline — Batch runner for the reader panel (#2170, CWQE Phase 6).
*
* Convenes the four-persona panel for a series, streaming per-persona progress
* to attached SSE clients. Mirrors editorialAnalysisRunner.js: one in-memory
* `runs` map keyed by `seriesId`, terminal-frame replay for late clients.
*
* This is the ONLY entry that spends LLM calls for the panel — it runs only from
* the explicit `POST /series/:id/editorial/panel/run` action (or an autopilot
* step), never at boot (AI-provider policy).
*/
import { createSseRunner } from '../../lib/sseUtils.js';
import { PANEL_PERSONAS } from '../../lib/editorial/panelDisagreement.js';
import { buildDigestForSeries } from './readerPanelDigest.js';
import { runPersona, finalizePanel, clearReaderPanel } from './readerPanel.js';
const runner = createSseRunner({ logLabel: 'reader panel' });
export function isReaderPanelActive(seriesId) {
return runner.isActive(seriesId);
}
export function attachClient(seriesId, res) {
return runner.attachClient(seriesId, res);
}
export function cancelReaderPanel(seriesId) {
return runner.cancel(seriesId);
}
/**
* Kick off the panel run. Returns the runId immediately; per-persona progress
* lands via SSE. Re-calling while a run is in flight resolves to the existing
* runId.
*/
export function startReaderPanel(seriesId, options = {}) {
return runner.start(seriesId, async ({ runId, record, broadcast }) => {
const digest = await buildDigestForSeries(seriesId);
if (!digest.issueCount) {
// No analyzable content — clear any stale consensus findings AND the stored
// snapshot a prior panel left, so the panel reflects "nothing to read" now.
await clearReaderPanel(seriesId, { runId });
broadcast({ type: 'complete', runId, personas: 0, empty: true, completedAt: new Date().toISOString() });
return;
}
broadcast({ type: 'start', runId, total: PANEL_PERSONAS.length, issueCount: digest.issueCount });
const responses = [];
let done = 0;
for (const persona of PANEL_PERSONAS) {
if (record.cancelRequested) break;
broadcast({ type: 'persona:start', persona: persona.id, label: persona.label, done, total: PANEL_PERSONAS.length });
const response = await runPersona(persona.id, digest, {
providerId: options.providerId,
model: options.model,
});
responses.push(response);
done += 1;
broadcast({ type: 'persona:complete', persona: persona.id, label: persona.label, done, total: PANEL_PERSONAS.length });
}
if (record.cancelRequested) {
broadcast({ type: 'canceled', runId, personas: responses.length, canceledAt: new Date().toISOString() });
return;
}
const snapshot = await finalizePanel(seriesId, digest, responses, { runId });
broadcast({
type: 'complete',
runId,
personas: responses.length,
consensus: snapshot.disagreements?.consensus?.length || 0,
seededFindings: snapshot.seededFindings || 0,
completedAt: new Date().toISOString(),
});
});
}
export const __testing = { runs: runner.runs };