The universal platform for autonomous bots.
Build, run, and evolve any bot — trading, AI agents, social, automation — from one cinematic command center. Self-hosted. Encrypted. Yours.
Your entire fleet.
One screen.
Every bot. Every metric. Every signal. The dashboard is the cockpit — adaptive, real-time, and ruthlessly clear. No menus to dig through. No tabs to babysit. Just the truth, surfaced instantly.
Adaptive UI
Detects each bot's type — trading, Telegram, AI agent — and surfaces only the metrics that matter for it.
Live Everything
Logs, P&L, system load, AI tokens, news. All streaming, color-coded, and inspectable in real time.
Zero Configuration
Paste Python, click Start. WATCH-DOG handles execution, restarts, logging, and analytics.
Code is a script.
A bot, alone, is a tool.
A fleet, orchestrated, is leverage.
Track your bots from
anywhere.
Use your phone or browser — just log in and monitor your live dashboard anytime. Bot status, P&L, logs, and announcements stream into the same web command center your desktop app talks to. One account, every device, zero re-configuration.
Web first
Open watchdogbot.cloud on any laptop or tablet, sign in, and your fleet is right there — no app install required to check in.
Mobile-ready
The same dashboard works on your phone. Pull up bot counts, recent logs, and announcements wherever you are.
Same source of truth
Your desktop app and the web dashboard read from the same encrypted store — what you see in one place matches what you see in the other.
Eight archetypes.
Infinite strategies.
Trading, Arbitrage, Prediction, Grid, DCA, Scalping, Telegram, Custom — eight bot shapes, one cockpit. Pick the archetype that fits your strategy. Spin it up in seconds.
Every tick. Every trade.
Audited live.
Watch every bot work. Health, uptime, errors, configuration, and a streaming log of every signal it sees and every decision it makes. P&L is parsed straight from the log — no SDK, no glue, no plumbing.
Log-Native Analytics
Trades, win rate, equity curve, profit factor — all parsed from your bot's stdout. Any Python script becomes instrumented.
Risk Guardrails
Per-bot daily-loss caps, trade-size limits, and run schedules — injected as env vars and enforced by the platform.
Process Isolation
Every bot runs as its own subprocess in its own sandbox. One crash never touches another bot. Auto-restart standard.
Bots that learn.
From your own data.
Train AI models on your bots' actual trade history — or upload your own datasets. Backtest or live-sync. Daily, every 25 trades, every 50. WATCH-DOG aggregates, analyzes, and ships back actionable performance reports. No cloud. No API key. Your data never leaves the machine.
Six Data Formats
CSV, JSON, JSONL, TXT, PDF, XLSX. Drop in historical files. Connect bots. Mix. Train.
Live Sync
New trades feed your model the moment they happen. Models compound knowledge with every fill.
Multi-Bot Models
Aggregate signal across an entire fleet. One model trained on five bots learns what one alone never could.
The system watches
itself.
CPU, RAM, AI tokens, financial news, bot uptime, logs-per-minute — all on one panel, all the time. The intelligence layer auto-classifies what each bot is doing — trading, messaging, scraping — purely from log content. A live fingerprint of your fleet.
Auto-Classification
Bots are fingerprinted from their log output: Exchange, Telegram, Discord, AI, Trading, Scraper, Notifier, Prediction.
Six Live Feeds
Reuters, Yahoo Finance, MarketWatch, Seeking Alpha, Investing.com, Google Finance — auto-rotating, 5-minute refresh.
System Vitals
Live CPU and RAM, with high-resource warnings at >75% / >85%. Token counters for every AI call.
Built with real traders.
For real traders.
The Community panel is where the WatchDog operators talk. Announcements, predictions market, fun-chat, support — and one-tap direct messages with anyone running a bot. Strategies travel faster between humans than between docs. So we built the chat in.
If it has an API,
it can run here.
Each bot type gets its own dashboard, its own metrics, its own intelligence. Detection happens automatically — from your code and your logs.
Engineered for the
few who run many.
Every detail is considered. Every system is observable. Every bot is yours.
P&L from log output.
Total trades, win rate, profit factor, equity curve, best and worst trades — extracted live from your bot's stdout via regex. No SDK required. Any Python bot becomes instrumented the moment you start it.
Logs are the source of truth.
Color-coded log levels. Smart inline tag detection — BUY, SELL, EXIT, SIGNAL, PNL, AI, ERROR, WARNING. Per-bot filtering, global activity feed, logs-per-minute meter. Stored in SQLite, streamed to the UI.
Process isolation.
Each bot is its own subprocess with its own working directory, its own log files, its own env. One crash never cascades.
Dynamic API connections.
Name a connection, paste your keys, and WATCH-DOG injects them as env vars. Coinbase, Kalshi, Anthropic templates ready.
Risk guardrails.
Per-bot daily-loss caps, trade-size limits, position-size limits, and time-window schedules. Injected as env vars and enforceable.
Built-in community chat — no third party.
Group channel, private DMs, presence, typing indicators, image and file sharing, persistent history. All on native FastAPI WebSockets, all on your own server. Zero Firebase. Zero Pusher. Zero data egress.
Seven themes.
Deep Space, Ocean, Cosmic, Matrix, Synthwave, Glacier, Light. Pre-paint loading. No flash.
Sign in. Start running.
Standard email-and-password with hashed credentials, optional Whop license verification, and per-bot UUID secrets for authenticated callbacks. Authentication happens locally — no third-party identity service ever sees your session.
- Hashed credentials stored in local SQLite
- Optional Whop integration for license-gated builds
- Per-bot secret tokens for authenticated API callbacks
- Sessions never leave the machine
Common questions.
What is WatchDog Bot?
WatchDog Bot is a desktop trading bot platform built for Python developers. You write bots in pure Python and paste them into the app — the platform handles auto-installing dependencies, surfacing errors with AI-powered fixes, real-time log streaming, and per-bot isolated execution. It supports Kalshi prediction markets, every major crypto exchange via CCXT, and any custom HTTP API.
Is there a free trial? Do I need a credit card?
Yes. Every new account gets a free trial automatically — no credit card required. You can build, run, and monitor bots during the trial with full feature access. The trial converts to a paid subscription only when you choose to subscribe.
What exchanges does WatchDog Bot support?
Kalshi (US-regulated prediction markets) has first-class built-in support. For crypto, the platform uses CCXT under the hood, which means Binance, Coinbase, Kraken, Bybit, OKX, KuCoin, Bitfinex, Bitstamp, and dozens more work out of the box. You can also add any HTTP API as a custom connection — useful for sportsbooks, alternative data providers, or in-house systems.
Do I need to know Python to use it?
Yes — bots are written in Python. If you can read and modify a 30-line Python script, you can use WatchDog Bot. We provide a library of copy-paste starter bots covering common patterns (momentum, DCA, market making, news-driven). You don't need framework knowledge — just standard Python.
Does it run on Mac and Windows?
Yes. WatchDog Bot ships as a signed desktop app for both Windows and macOS. The installer is around 200 MB and includes a bundled Python runtime — no separate Python install needed. Linux support is on the roadmap but not yet available.
How does AI Fix work?
When a bot crashes with an unhandled exception, the dashboard shows a "Fix with AI" button. Clicking it sends your bot's code + the traceback + recent logs to Claude (Anthropic's model), which proposes a code diff with explanation. You review the diff and apply it with one click. See the full AI Fix guide for details on what gets sent and when to trust the suggestion.
Can I run multiple bots at the same time?
Yes — that's the core design. Each bot gets its own isolated Python venv, so different bots can use different package versions without conflict. The dashboard shows every bot's status, logs, and metrics at once. No artificial cap on the number of bots per account.
Where are my API keys stored?
API keys live in an encrypted local connection store and (optionally) sync to your account in Supabase. Bots never see plaintext keys in code — they access credentials via the built-in wd.connection("ExchangeName") SDK call. AI Fix and our backend never see your API keys, even when debugging crashes.
Does WatchDog Bot take a percentage of my trades?
No. We never touch your funds or trades. WatchDog Bot is a software subscription — your API keys, your account, your trades, your P&L. We have no order routing, no custody, no rev-share. The only money changing hands is your subscription fee.
What if my bot keeps crashing?
The platform handles the most common crash modes automatically: missing dependencies are auto-installed and the bot retries; common Python errors get friendly messages with line numbers and fix hints. For everything else, AI Fix usually proposes a working fix in seconds. See our guide on the 7 most common crash causes for a deeper walk-through.
Can I backtest a strategy in WatchDog Bot?
Basic backtesting is supported — you can replay historical data through a bot script. Rich backtesting metrics (Sharpe, Sortino, walk-forward optimization) are on the roadmap but not yet at the level of Freqtrade or vectorbt. For now, many users backtest in Freqtrade and run live bots in WatchDog Bot. See our backtesting guide for the full picture.
How do I cancel my subscription?
One click in account settings. No phone calls, no retention flows. If you cancel mid-month, you keep access until the end of your billing period.
Run your fleet
like an institution.
A single download. A lifetime license. Every bot you've ever wanted to run, on infrastructure you own. Self-hosted. Local-first. Built for the few.