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 { useState, useEffect, useRef, useCallback } from 'react';
import { Terminal, RefreshCw } from 'lucide-react';
import * as api from '../../services/api';
import socket from '../../services/socket';
const MAX_LIVE_LINES = 500;
export default function OutputTab({ agent }) {
const [output, setOutput] = useState('');
const [agentId, setAgentId] = useState(null);
const bottomRef = useRef(null);
// Monotonic request counter: only the most recently issued fetch may write
// state. An effect-local `cancelled` flag can't cover this, because the
// refresh button fires outside the effect — and comparing agent ids alone
// wouldn't either, since two fetches for the *same* agent (a run starting,
// or a second refresh click) can still resolve out of order and repaint an
// older run's output.
const requestSeqRef = useRef(0);
const loadOutput = useCallback((id) => {
const seq = ++requestSeqRef.current;
api.getFeatureAgentOutput(id).then(data => {
if (seq !== requestSeqRef.current || !data) return;
setOutput(data.output || '');
setAgentId(data.agentId);
}).catch(() => {});
}, []);
// Drop the previous agent's output the moment the selection changes —
// otherwise it lingers until the new fetch lands and reads as though the
// newly selected agent had already produced it. Keyed on `agent.id` alone so
// a run starting (`currentAgentId` change) doesn't wipe streamed output.
useEffect(() => {
setOutput('');
setAgentId(null);
}, [agent.id]);
useEffect(() => {
loadOutput(agent.id);
const handler = (data) => {
if (data.agentId === agent.currentAgentId && data.line) {
setOutput(prev => {
// The fetched snapshot may not end in a newline; keep the first
// streamed line from gluing onto its last line.
const sep = prev && !prev.endsWith('\n') ? '\n' : '';
const next = `${prev}${sep}${data.line}\n`;
// Keep the last MAX_LIVE_LINES lines (the trailing newline leaves one
// empty entry after the split) so a long run can't grow unbounded.
const lines = next.split('\n');
return lines.length > MAX_LIVE_LINES + 1 ? lines.slice(-(MAX_LIVE_LINES + 1)).join('\n') : next;
});
}
};
socket.on('cos:agent:output', handler);
return () => {
socket.off('cos:agent:output', handler);
};
}, [agent.id, agent.currentAgentId, loadOutput]);
useEffect(() => {
bottomRef.current?.scrollIntoView({ behavior: 'smooth' });
}, [output]);
if (!agent.currentAgentId && !output) {
return (
<div className="text-center py-16">
<Terminal size={48} className="mx-auto text-gray-600 mb-4" />
<h3 className="text-lg font-medium text-gray-400 mb-2">No Active Run</h3>
<p className="text-sm text-gray-600">Output will appear here when the agent is running.</p>
</div>
);
}
return (
<div className="space-y-3">
<div className="flex items-center justify-between">
<div className="text-sm text-gray-500">
{agent.currentAgentId ? (
<span className="text-port-success">Running: {agent.currentAgentId}</span>
) : (
<span>Last run output{agentId ? ` (${agentId})` : ''}</span>
)}
</div>
<button
onClick={() => loadOutput(agent.id)}
aria-label="Refresh output"
className="min-h-[44px] min-w-[44px] inline-flex items-center justify-center p-1.5 text-gray-400 hover:text-white hover:bg-port-border/50 rounded transition-colors"
>
<RefreshCw size={14} />
</button>
</div>
<div className="bg-port-bg border border-port-border rounded-lg p-4 font-mono text-xs text-gray-300 max-h-[600px] overflow-auto whitespace-pre-wrap">
{output || 'No output yet...'}
<div ref={bottomRef} />
</div>
</div>
);
}