stonkbro — AI-Powered Options Trading Copilot

Features

20+ production features. Zero mock data.

Every feature pulls live market data from Tradier and persists real positions in Supabase. Tradier provides real Greeks on every contract.

🔍

Discovery Dashboard

Ranked watchlist with custom scoring: volume ratio, momentum, SMA positioning, 52-week range. High scores = explosive potential.

Live Data

🌏

Sector Discovery

Browse 8 curated sectors (AI Infra, Quantum, Nuclear, Space, Fintech, Biotech, EV, AI Software) with 70+ pre-loaded tickers. Expandable card grid with live quotes.

70+ Tickers

🔥

Explosive Finder

Claude AI analyzes each sector to surface the highest-conviction 10x potential stocks, with specific entry strategies and risk reasoning.

AI-Powered

☀

Daily Recommendations

AI-generated daily stock picks across 3 themes (momentum, value, contrarian) — cached by cron at market open, refreshable on demand. Collapsible theme cards with next-refresh countdown.

Cron-Cached

📊

PMCC Scanner

Scans real options chains for optimal LEAPS + short call pairings. Grades A/B/C by capital efficiency and annualized return. Real delta from Tradier.

Options Chains

🎯

PMCC Picks

Auto-scan sectors for the best PMCC setups ranked by monthly income and capital efficiency. Combines sector discovery with options analysis.

Auto-Scan

🧠

Deep Research Engine

Claude AI analyzes up to 20 symbols: market context, technicals, and generates specific trade suggestions with strike, expiry, and reasoning.

AI-Powered

💼

Position Tracking

Log multi-leg trades (PMCC, CC, CSP, Wheel). Track each leg with strike, expiry, entry price. Lifecycle: active → rolled → closed.

Full CRUD

🛑

Trailing Stop System

Set % trailing stops on positions. App tracks peak price automatically, calculates drawdown, and flags when stop triggers. Color-coded risk view.

Risk Management

💰

Portfolio (Live P&L)

Real P&L from Supabase positions + live Tradier quotes. Summary cards, per-position breakdown, gain/drawdown tracking, % from peak visualization.

Real-Time

⚠

Trade Signals

Real-time alerts: CLOSE at 50%+ profit, ROLL at 21 DTE, WARNING when price nears strike. Urgency-sorted: HIGH / MEDIUM / LOW.

Alerts

📈

Covered Call Optimizer

Find best CC strikes for your holdings. Filters 3-10% OTM, 20-45 DTE. Grid view with annualized return and probability OTM.

Options Chains

🔄

Wheel Visualizer

Timeline of the full wheel cycle: Sell Put → Assigned → Sell Call → Called Away. Premium per step, cumulative income.

Strategy View

💵

Income Dashboard

Track your premium income: starting capital, total collected, yield %, monthly breakdown chart, annualized projections.

Analytics

📋

Watchlist Management

Create named watchlists, add/remove tickers, set defaults. Your default watchlist drives Discovery. Persisted with RLS.

Supabase

📧

Automated Cron Alerts

Vercel cron runs multiple times daily on market days. Checks all positions, generates daily recommendations, sends email briefings via Resend.

Automated

📊

Admin Panel + Token Tracking

Admin-gated settings panel for switching AI providers and viewing real-time token usage per feature. Backed by ai_token_usage table with full audit trail.

Admin

🔐

Google Auth

One-click Google login via Supabase Auth. All routes protected. Row-level security ensures data isolation per user.

Secure

Database

10 tables, full RLS

Every user-scoped table has row-level security. Global config tables have public read, admin write. Users only see their own data.

Table	Purpose	Key Fields
`watchlists`	Named watchlists per user	name, is_default, user_id
`watchlist_items`	Tickers in each watchlist	symbol, watchlist_id (unique pair)
`positions`	Options/stock positions	symbol, strategy, status, trailing_stop_pct, peak_price
`position_legs`	Individual legs of multi-leg trades	type, strike, expiry, entry_price, qty
`research_reports`	Claude AI analysis results	symbols_analyzed[], report (markdown)
`trade_suggestions`	Specific trade recommendations	symbol, strategy, strike, premium, reasoning
`daily_recommendations`	Cron-cached daily AI stock picks	date, theme, picks (jsonb), generated_at
`user_settings`	Preferences, admin flag & account defaults	starting_cash, alert_email, is_admin
`app_config`	Global AI provider settings	key, value — public read, admin write
`ai_token_usage`	Token audit trail per AI call	feature, user_id, input_tokens, output_tokens, model

API

12 backend routes

Server-side endpoints for market data, AI analysis, signals, and automated alerts.

Endpoint	Method	Description
`/api/research`	POST	Run Claude AI deep research on up to 20 symbols. Returns markdown report + trade suggestions.
`/api/recommendations`	GET	Daily AI stock picks across 3 themes. Served from cache; force-refresh via query param.
`/api/explosive`	POST	Claude AI explosive stock finder for a given sector. Returns high-conviction picks with entry strategies.
`/api/options?symbol=X`	GET	Fetch PMCC setups for a symbol from live Tradier options chains with real Greeks.
`/api/pmcc-scan`	POST	Batch PMCC scan across a list of tickers. Returns ranked setups by monthly income.
`/api/scanner`	GET	Discovery scoring engine — ranks watchlist tickers by explosive potential score.
`/api/signals`	GET	Check active positions for roll/close/profit alerts. Returns urgency-sorted signal list.
`/api/movers`	GET	Top market movers and unusual volume across watchlist symbols.
`/api/earnings`	GET	Upcoming earnings dates for watchlist symbols — risk flag for options positions.
`/api/cron`	GET	Automated multi-purpose cron (Vercel cron). Runs position alerts, daily recs refresh, email briefings.
`/api/admin`	GET	Admin stats — token usage by feature, current AI provider config. Admin-gated.
`/api/admin/provider`	POST	Toggle AI provider (Anthropic / OpenAI) and default model. Updates `app_config` table.

History

From zero to 20+ features

Built at mass speed with Claude Code.

0.10

Sector Discovery, Explosive Finder, Trailing Stops, PMCC Picks

8 curated sectors with 70+ tickers, Claude AI explosive stock finder, trailing stop system with peak tracking, PMCC auto-scan ranked by income.

0.9

Tradier Integration + Mock Fallback

Replaced Yahoo Finance with Tradier for real Greeks (delta, gamma, theta, vega, IV). Automatic mock fallback when API key not set.

0.8

All Real Data: No More Mocks

Every page wired to live market data + Supabase. Zero hardcoded data remaining.

0.7

Full App: Watchlists, Positions, Research

Watchlist management, position tracking with legs, Claude AI research engine, options suggestions. 7 Supabase tables.

0.6

Cron Alerts & Trade Signals

Vercel cron jobs 3x daily. Signal engine for profit targets, DTE warnings, strike breaches. Email via Resend.

0.5

Live PMCC Scanner

Real options chain scanning. LEAPS + short call pairing with A/B/C grades. Yahoo Finance market data layer.

0.4

Auth, Covered Calls, Wheel, Income

Google OAuth, CC Optimizer, Wheel Visualizer, Income Dashboard. Route groups for auth/unauth.

0.3

Discovery, Ticker Detail, PMCC, Portfolio

Core feature pages with technical indicators, scoring algorithm, and PMCC analyzer.

0.2

App Shell & Navigation

Layout, bottom nav, empty state pages. Light stone theme.

0.1

Project Init

Next.js + Supabase + Vercel scaffolding. Project conventions established.

Roadmap

What's next

Current P2 backlog — items Claude can pick up and implement autonomously once promoted to P1.

Email Activation

Configure Resend API key to activate automated email alert briefings

Paper Trading Mode

Simulate trades with $20k virtual capital, track hypothetical performance

Research → Position

Accept an AI suggestion and auto-create the position with one click

Broker Integration

Connect to Tradier or IBKR for one-click order execution

Rules Engine

Define auto-rolling, profit-taking, and defense rules

Risk Dashboard

Correlation matrix, sector exposure, portfolio delta

Backtesting

Test scoring model and strategies against historical data

PWA + Push

Install to home screen with native push notifications

Developer

Autonomous Backlog Mode

Claude Code can pick items from the backlog, implement them, commit, and loop — without supervision. Karthik reviews diffs and promotes items to approve them.

/project:generate-backlog

Analyze the codebase for bugs, gaps, and ideas. Adds confirmed items to BACKLOG.md P2, and feature ideas to IDEAS.md for review.

/project:list-backlog

Print a clean count of P1 / P2 / In Progress / Done items and list each title. Read-only, no file changes.

/project:work-backlog

Pick the next P1 item (or P2 if P1 empty), implement it fully, run build, commit on success, and loop to the next item.

./run-backlog.sh

Shell loop that re-invokes /project:work-backlog until Claude writes a status file — DONE, BLOCKED, NEEDS_INPUT, or LOOP_DETECTED.

Auto-approved (Claude just does it)

Bug fixes with clear reproduction in the task description

Missing error handling, null checks, TypeScript type fixes

Console.log cleanup and minor copy/label fixes

Test additions for existing behavior

Performance improvements with no behavior change

Requires stopping (NEEDS_INPUT)

New pages, routes, or major UI sections not in the task

New external API or data source integrations

Changes to core business logic or calculations

Anything requiring new environment variables

Deleting or renaming existing database tables or routes

Exit status written to .claude-status

DONE: all items complete BLOCKED: [task] — [reason] NEEDS_INPUT: [task] — [what is needed] LOOP_DETECTED: [what is repeating]