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
import { defineConfig, devices } from "@playwright/test";
/**
* Playwright config for the AiSOC web app.
*
* Three projects live here:
*
* 1. ``screencast`` (Phase 4.3) — records the 90-second product demo
* against a deployed URL configured by ``AISOC_SCREENCAST_URL``.
* Defaults to the local console, same as the ``screenshots`` project —
* the two used to disagree, so an unset env var silently recorded
* against a different deployment than the screenshots did.
*
* 2. ``visual`` (Phase 4.7) — Storybook-driven visual regression. The
* ``webServer`` block boots a static `http-server` over
* ``storybook-static/`` on port 6007, which the visual spec hits
* via ``/iframe.html?id=...``. Run ``pnpm build-storybook`` first,
* otherwise the webServer has nothing to serve. Baselines live next
* to the spec and are committed to git.
*
* 3. ``journey`` (Phase 4.8) — buyer journey E2E. The webServer block
* boots ``next dev`` on port 3100 (so it doesn't clash with a
* developer's local 3000) and the spec stubs every backend call via
* ``page.route()`` so the run is hermetic.
*
* Project selection is driven by ``PLAYWRIGHT_PROJECT`` so the
* webServer block doesn't try to boot a static Storybook for the
* journey run (or vice versa). When no project is selected (e.g. local
* ``playwright test`` with no flags), neither webServer is started —
* the screencast spec is the only one that doesn't need one.
*/
const PROJECT = (process.env.PLAYWRIGHT_PROJECT ?? "").toLowerCase();
const IS_VISUAL = PROJECT === "visual";
const IS_JOURNEY = PROJECT === "journey";
const IS_SCREENSHOTS = PROJECT === "screenshots";
// 3.5+ — assertions against the live demo stack, as opposed to the hermetic
// journey project and the screenshot recorders.
const IS_DEMO_STACK = PROJECT === "demo-stack";
export default defineConfig({
testDir: "./e2e",
// Each test gets up to two minutes; the screencast spec itself
// tops out around 90s of timed shots plus boot overhead. The visual
// spec is slower (it walks every Storybook story) so we extend the
// budget only when running that project.
timeout: IS_VISUAL ? 600_000 : 120_000,
expect: { timeout: 15_000 },
fullyParallel: false,
forbidOnly: !!process.env.CI,
retries: process.env.CI ? 0 : 0,
workers: 1,
reporter: process.env.CI ? "line" : "list",
// Each project gets its own webServer (or none for the screencast,
// which targets an externally-deployed URL).
...(IS_VISUAL && {
webServer: {
command:
"pnpm exec http-server storybook-static -p 6007 -s --cors --silent",
url: "http://127.0.0.1:6007/index.json",
reuseExistingServer: !process.env.CI,
timeout: 60_000,
},
}),
...(IS_JOURNEY && {
webServer: {
// Boot Next on 3100 so a developer's `pnpm dev` on 3000 doesn't
// collide with this run. `--turbo` keeps the boot under 20s on
// modern machines. Bind to `localhost` (not `127.0.0.1`) so
// Next 16's allowedDevOrigins guard doesn't block HMR / asset
// requests with "Blocked cross-origin request to Next.js dev
// resource /_next/webpack-hmr from 127.0.0.1".
command: "pnpm exec next dev -p 3100 --turbo",
url: "http://localhost:3100",
reuseExistingServer: !process.env.CI,
timeout: 180_000,
env: {
// The buyer-journey spec stubs every backend route via
// `page.route()`. Leaving `NEXT_PUBLIC_API_URL` empty means
// all API calls go same-origin and our route matcher catches
// every one of them.
NEXT_PUBLIC_API_URL: "",
},
},
}),
use: {
baseURL: IS_VISUAL
? "http://127.0.0.1:6007"
: IS_DEMO_STACK
? (process.env.AISOC_DEMO_WEB_URL ?? "http://localhost:3000")
: IS_JOURNEY
? "http://localhost:3100"
: IS_SCREENSHOTS
? (process.env.AISOC_SCREENCAST_URL ?? "http://localhost:3000")
: (process.env.AISOC_SCREENCAST_URL ?? "http://localhost:3000"),
trace: IS_JOURNEY ? "retain-on-failure" : "off",
screenshot: IS_JOURNEY ? "only-on-failure" : "off",
video: "off",
viewport: { width: 1920, height: 1080 },
actionTimeout: 10_000,
},
projects: [
{
name: "screencast",
testMatch: /demo\/screencast\.spec\.ts$/,
use: {
...devices["Desktop Chrome"],
viewport: { width: 1920, height: 1080 },
deviceScaleFactor: 1,
// The recorder writes a webm into the test output dir. We
// override the default codec config so the encoder is
// deterministic across runner generations.
video: {
mode: "on",
size: { width: 1920, height: 1080 },
},
},
},
{
name: "visual",
testMatch: /visual\/visual\.spec\.ts$/,
use: {
...devices["Desktop Chrome"],
// Force 1× DPR so screenshots have a deterministic pixel count
// regardless of the developer's Retina display vs. the CI VM.
deviceScaleFactor: 1,
viewport: { width: 1440, height: 900 },
},
},
{
// 3.5+ — assertions against the stack `pnpm aisoc:demo` starts. Kept
// separate from `journey`, which stubs the network: these two answer
// different questions and merging them would let a hermetic pass
// stand in for a live one.
name: "demo-stack",
testMatch: /demo-stack\/.*\.spec\.ts$/,
use: {
...devices["Desktop Chrome"],
viewport: { width: 1440, height: 900 },
},
},
{
name: "journey",
testMatch: /journeys\/.*\.spec\.ts$/,
use: {
...devices["Desktop Chrome"],
viewport: { width: 1440, height: 900 },
},
},
{
// Phase 2 quick-win on-ramp: README hero screenshots. Runs
// against a seeded local demo (`pnpm aisoc:demo`) or any
// deployed URL pointed at by `AISOC_SCREENCAST_URL`. Output
// viewport is 1440x900 so the captured PNGs land in the same
// 1.6:1 aspect ratio the README image grid expects.
name: "screenshots",
testMatch: /screenshots\/.*\.spec\.ts$/,
use: {
...devices["Desktop Chrome"],
deviceScaleFactor: 1,
viewport: { width: 1440, height: 900 },
},
},
],
});