Watchtower MCP
Ekonoski/watchtower-mcp
Scanner de trading ao vivo integrado ao Claude via MCP
Ver no GitHubServidor MCP que expõe a plataforma de busca de ações Watchtower para Claude e outros clientes MCP. Oferece dashboard ao vivo com scanners intraday, análise de notícias com IA, alertas em tempo real e histórico de performance de sinais de trading.
Instalação
npx convez-ai Ekonoski/watchtower-mcp1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Buscar configurações de day-trading com catalistas de notícias em tempo real
- Monitorar sinais técnicos (GAP_AND_GO, VWAP_BREAKOUT, FLUSH_REVERSAL) no mercado americano
- Analisar sentimento de mercado via dados de X/redes sociais e Grok
- Consultar performance histórica de tipos de sinais para validação de estratégias
Veredito

Ferramenta especializada para traders que querem integrar análise de mercado em tempo real ao Claude. Bem arquitetada com scanners automáticos, persistência de dados e alertas gated, mas exige conhecimento profundo de trading e configuração infraestrutural (Railway, Polygon, xAI, Supabase).
README
A tradução para o português deste README ainda está na fila.
Watchtower MCP — Live Scanner Service
The always-on Railway service for the Watchtower trading platform:
- Live dashboard at
/dashboard— the primary interface. Auto-surfaces day-trading setups across the full US market with news catalysts, X/social sentiment, price/volume/% change, and browser pop-up alerts. - MCP server at
/mcp— exposes every screen as a tool for Grok, Claude, and other MCP clients (Bearer-token auth). - Background scheduler — scans every 5 minutes during market hours, every 15 minutes pre-market, plus daily screen sweeps; persists every scan and sends gated email alerts.
Dashboard
Open https://<your-railway-domain>/dashboard.
- Scanner tab — every intraday signal from the latest scan (GAP_AND_GO, VWAP_BREAKOUT, FLUSH_REVERSAL, breakdowns, volume surges…) with score, price, % change, gap %, volume pace, VWAP side, and sector. Filter by direction/score/ticker, sort any column, click a row for a live drill-down with X sentiment.
- News Catalysts tab — Grok-classified market-moving news (earnings, FDA, M&A, analyst actions…) cross-referenced with live technicals into BUY/WATCH/AVOID calls. Off-radar tickers flagged.
- Performance tab — win rates and return paths (d1→d90) for every signal
type ever fired, straight from
alert_log. - Watchlist tab — manage the tickers that are always included in scans.
- 🔔 Alerts — browser pop-up notifications when new signals (score ≥ 55) or high-impact news land. The page polls every 30 s; scans run every 5 min.
Dashboard auth
Set DASHBOARD_PASSWORD in Railway variables (falls back to
MCP_AUTH_TOKEN). Session lives in an HttpOnly cookie for 30 days.
Scan → alert flow
every 5 min (mkt hours) every scan gated
┌──────────────────┐ ┌───────────────────────┐ ┌──────────────┐
│ intraday screen │ → │ scan_snapshots (DB) │ → │ email alert │
│ news scan (Grok) │ │ + alert_log tracking │ │ fresh signals│
│ social pulse (X) │ │ → dashboard reads this │ │ only, 60-min │
└──────────────────┘ └───────────────────────┘ │ per-ticker │
│ cooldown │
└──────────────┘
Email gating knobs (Railway variables):
| Variable | Default | Meaning |
|---|---|---|
ALERT_EMAIL_MIN_SCORE | 55 | Min intraday score to count as email-worthy |
ALERT_EMAIL_COOLDOWN_MIN | 60 | Per-ticker email cooldown |
ALERT_EMAIL_HEARTBEAT_MIN | 60 | Max minutes between emails during scans |
EMAIL_EVERY_SCAN | unset | true restores email-on-every-scan |
POLYGON_DATA_DELAY_MIN | 15 | Snapshot delay; set 0 on a real-time plan |
Environment variables
See RAILWAY_ENV_VARS.md for the full list (Polygon, xAI, Supabase, FMP,
email transport, MCP auth). Never paste actual keys into any file in this
repo.
Migrations
migrations/*.sql — apply in order against the Supabase project. The
scan_snapshots table is also auto-created on first write.
Deployment
Railway: railway.toml (healthcheck /health) / Procfile
(uvicorn server:app). After changing variables, redeploy.