▶ If the player above doesn't load, watch / download the demo video directly.
Athena is a personal command center — a single retro-terminal "starship bridge" web app that runs two very different systems from one interface:
- A website-production pipeline that finds leads, builds real multi-page sites with an AI agent, reviews them, and deploys them — mostly hands-off.
- MOTHER, a self-contained cybersecurity learning lab for practicing red/blue-team skills against your own machines.
This repo (mother-deck) is the complete build with both halves. The
website-pipeline half is also published on its own, with the security deck
removed, as athena-web-builder.
Secrets, API keys, client data, and finished client sites have been removed; the code and lab tooling are intact. It ships with localhost defaults — you bring your own keys.
Authorized use only. The MOTHER deck includes offensive security tooling intended only for use against systems you own or are explicitly authorized to test (your own lab VMs, CTF targets, engagements with written permission). It exists for educational and defensive purposes. Do not point it at systems you do not own or have permission to test.
An agent-driven assembly line that turns a lead into a shipped website. The
control bridge (frontend/) drives it, and a Node/Express backend
(backend/src/pipeline) runs the work:
leads → build → review → deploy
- Leads are stored and queued.
- Build shells out to an external AI agent (the OpenClaw CLI) which scaffolds a real multi-page site from a template.
- Review collects the generated sites into an inbox for approval.
- Deploy optionally ships to Vercel and reports back over Telegram / SMS.
An auto-scheduler ("the circuit") can run this whole loop on a cadence. The backend orchestrates the agent — it does not contain the LLM itself.
The bridge UI has sections for chat (COMMS), automations (OPERATIONS), the build lane (PIPELINE), a catalog of ~150 mined UI patterns (SHOWCASE), a review queue (INBOX), a live event stream (MISSION LOG), and a notes browser (MEMORY).
A self-contained security lab under frontend/public/cyber/, backed by three
server surfaces. Everything defaults to off and localhost-only:
backend/src/routes/reconRoutes.ts— passive OSINT aggregator. Orchestrates standard, read-only reconnaissance tools (crt.sh, theHarvester, subfinder, amass, and Shodan's pre-scanned database), validates every domain against an injection guard, keeps secrets out of argv/history, and correlates the results into one view. Passive only — it never touches the target.backend/src/pty.ts— web terminal relay. A browser terminal (xterm.js) wired to a real shell over WebSocket, with ssh / WSL / Docker / local / VM backends. This is remote code execution by design, so it stays disabled unlessATHENA_TERMINAL_ENABLED=1, is bound to localhost, and is gated by a per-session token.backend/src/learn.ts— guided-demo planner. A local tutor that scripts walkthroughs of the deck's tools.
The deck's static pages cover a recon map, blue-team log analysis, a command reference, payload / lab tooling, kill-chain and engagement trackers, and a session recorder — all using synthetic lab data.
Browser (React bridge) ──HTTP/WS──► Express backend ──spawns──► OpenClaw agent CLI
:5173 :3001 (build worker)
│
Vercel deploy · Telegram/Twilio reports · MOTHER deck surfaces
npm install
cp .env.example backend/.env.local # fill in your values
npm run dev # backend :3001, frontend :5173The website pipeline runs with localhost defaults. Deck features stay dark until
you opt into them via env (ATHENA_TERMINAL_ENABLED, SHODAN_API_KEY, VM
settings). See .env.example for every setting, and
SECURITY-HARDENING-REPORT.md for the backend hardening notes (loopback binding,
injection guards, SSRF protection, secret hygiene).
MIT — see LICENSE.