Open-source AI Security Operations Center: alert fusion, LLM-agent triage, MITRE ATT&CK investigation, and a replayable decision ledger for every agent step. Self-hostable, runs with no API keys, MIT licensed. Ships an MCP server for Claude, Cursor and Continue.
Python agentic-aiai-securityalert-triageblue-teamcybersecuritydetection-engineeringincident-responselanggraphllm-agentsmcpmitre-attackmodel-context-protocolpurple-teamsecurity-operationsself-hostedsiemsoarsocthreat-detectionthreat-intelligence
'use client';
import { useState } from 'react';
import Link from 'next/link';
import useSWR from 'swr';
import toast from 'react-hot-toast';
import { onboardingApi, type OnboardingStatus, type SampleDataResult } from '@/lib/api';
import { ModelPlacement } from './ModelPlacement';
import { clsx } from 'clsx';
/**
* What a new tenant still has to do, and a way to see the product work
* before they have credentials for anything.
*
* The gap this fills
* ------------------
* A brand-new operator signed in and landed on `/dashboard`: every tile
* zero, every panel an honest empty state, and nothing saying what to do
* next. The empty states were right — "0 connected sources" is true — but
* true and *useful* are different things, and none of them is a button.
*
* Two paths, deliberately
* -----------------------
* **Connect something real** is the primary one and is listed first,
* because that is what the product is for.
*
* **Load sample data** is the escape hatch for the evaluator who is three
* approvals away from an API key for their EDR and wants to know whether
* this is worth pursuing. It pushes five scenarios through the same
* ingest endpoint a real connector uses, so what they end up looking at
* has genuinely been normalised, correlated and triaged — a console full
* of inserted rows would look identical whether the pipeline works or is
* completely broken.
*
* What it will not do
* -------------------
* Sample data does not mark setup complete. Somebody who has only looked
* at samples still has nothing connected, and telling them otherwise
* would be exactly the kind of flattering fiction this product spends
* most of its effort avoiding.
*/
export function SetupChecklist({ compact = false }: { compact?: boolean }) {
const { data, error, isLoading, mutate } = useSWR<OnboardingStatus>(
'onboarding-status',
onboardingApi.status,
{ refreshInterval: 15_000 },
);
const [loading, setLoading] = useState(false);
const [result, setResult] = useState<SampleDataResult | null>(null);
const loadSamples = async () => {
setLoading(true);
try {
const outcome = await onboardingApi.loadSampleData();
setResult(outcome);
toast.success(
`${outcome.accepted} sample events accepted. They become alerts in a few seconds.`,
);
// Revalidate rather than optimistically marking the step done: the
// events are in the pipeline, not yet alerts, and claiming
// otherwise is the exact shape of dishonesty this avoids.
void mutate();
} catch (err) {
const message = err instanceof Error ? err.message : String(err);
// Surfaced verbatim. A first run is the worst possible moment to
// replace a specific cause with "something went wrong".
toast.error(message, { duration: 10_000 });
} finally {
setLoading(false);
}
};
if (error) {
return (
<div className="rounded-lg border border-amber-500/30 bg-amber-500/5 p-4 text-sm">
<p className="font-medium text-amber-200">Setup status unavailable</p>
<p className="mt-1 text-slate-400">
The console could not read what is set up. Treat this as unknown rather than as a
tenant with nothing configured.
</p>
</div>
);
}
if (isLoading || !data) {
return <div className="h-32 animate-pulse rounded-lg bg-slate-800/40" />;
}
// `try` and `model` are both optional: sample data is an alternative to a real
// connector, and the model step is informational because a local model ships
// and works. Counting either would show a tenant that is genuinely finished as
// having work left.
const remaining = data.steps.filter(
(s) => !s.done && s.key !== 'try' && s.key !== 'model',
).length;
return (
<section
aria-labelledby="setup-checklist-heading"
className="rounded-xl border border-slate-700/60 bg-slate-900/40 p-5"
>
<div className="flex items-start justify-between gap-4">
<div>
<h2 id="setup-checklist-heading" className="text-lg font-semibold text-slate-100">
{remaining === 0 ? 'Setup complete' : 'Finish setting up'}
</h2>
<p className="mt-1 text-sm text-slate-400">
{remaining === 0
? 'A source is connected and alerts are arriving from it.'
: `${remaining} step${remaining === 1 ? '' : 's'} left before AiSOC is working on your own data.`}
</p>
</div>
</div>
<ol className="mt-5 space-y-3">
{data.steps.map((step) => (
<li
key={step.key}
className={clsx(
'flex gap-3 rounded-lg border p-3',
step.done
? 'border-emerald-500/25 bg-emerald-500/5'
: 'border-slate-700/60 bg-slate-800/30',
)}
>
<span
aria-hidden
className={clsx(
'mt-0.5 flex h-5 w-5 shrink-0 items-center justify-center rounded-full text-xs font-bold',
step.done ? 'bg-emerald-500/20 text-emerald-300' : 'bg-slate-700 text-slate-400',
)}
>
{step.done ? '✓' : '·'}
</span>
<div className="min-w-0 flex-1">
<div className="flex flex-wrap items-center gap-2">
<span className="font-medium text-slate-100">{step.label}</span>
<span className="sr-only">{step.done ? 'done' : 'not done yet'}</span>
{step.detail && (
<span className="text-xs text-slate-500">{step.detail}</span>
)}
</div>
{!compact && <p className="mt-1 text-sm text-slate-400">{step.why}</p>}
{step.key === 'model' && <ModelPlacement />}
{step.key === 'try' && !step.done && (
<button
type="button"
onClick={loadSamples}
disabled={loading}
className="mt-2 rounded-md border border-slate-600 px-3 py-1.5 text-sm text-slate-200 transition hover:border-slate-400 hover:bg-slate-800 disabled:opacity-50"
>
{loading ? 'Pushing events through the pipeline…' : 'Load sample data'}
</button>
)}
{step.href && !step.done && step.key !== 'try' && (
<Link
href={step.href}
className="mt-2 inline-block rounded-md bg-indigo-600 px-3 py-1.5 text-sm font-medium text-white transition hover:bg-indigo-500"
>
{step.key === 'connector' ? 'Connect a source' : 'Open'}
</Link>
)}
</div>
</li>
))}
</ol>
{result && (
<div className="mt-4 rounded-lg border border-slate-700/60 bg-slate-800/30 p-4">
<p className="text-sm text-slate-300">{result.note}</p>
<ul className="mt-3 space-y-1.5">
{result.scenarios.map((scenario) => (
<li key={scenario.key} className="text-sm">
<span className="text-slate-200">{scenario.title}</span>
<span className="ml-2 text-xs uppercase tracking-wide text-slate-500">
{scenario.severity}
</span>
<p className="text-xs text-slate-500">{scenario.why}</p>
</li>
))}
</ul>
<Link
href="/alerts"
className="mt-3 inline-block text-sm text-indigo-400 hover:text-indigo-300"
>
Watch them arrive in Alerts →
</Link>
</div>
)}
{data.sample_data_loaded && (
<p className="mt-4 text-xs text-slate-500">
Sample alerts are attributed to "AiSOC" in the source column, so you can tell
them from real telemetry anywhere in the console. Connecting a real source is still
the next step.
</p>
)}
</section>
);
}