Robinhood Pro

Confluence Pro MCP

The Pro tier includes a standard MCP server (not tied to one editor) that streams graded alerts, gamma rails, and a Robinhood execution checklist. Use it in Cursor, Claude Desktop, or VS Code with an MCP extension. Pair with Robinhood’s official MCP for quotes and orders when your client supports it. Advisory only — you confirm every trade.

Easiest: hosted connector (no install)

On a paid Claude plan (Pro, Max, Team, or Enterprise), Claude supports custom connectors — so you skip Python and config files entirely. Add one URL with your Pro key:

https://mcp.0dteconfluence.com/mcp?key=YOUR_cfl_KEY

In Claude: Settings → Connectors → Add custom connector → paste the URL → the Confluence tools appear in every chat. Your key scopes the connection to your account. Free Claude plans don’t include custom connectors — use the local install below instead.

Compatible clients

• Cursor — .cursor/mcp.json (recommended; Robinhood MCP URL works here)
• Claude Desktop — claude_desktop_config.json
• VS Code — MCP-capable extension (e.g. GitHub Copilot MCP); same server path + env
• Robinhood app — no MCP required; use #alerts-pro Discord backup + playbook

What you get

• Pro API key (cfl_…) on the success page after checkout
• 25+ MCP tools — alerts, rails, tape, morning desk, market clock, IV skew, pin, GEX, expected move, RV regime, SPY/QQQ divergence, replay, tradeability gate, Robinhood entry plan, brackets, plus Quant/Macro (path risk, CVD, VRP, index health, macro events, daily structure)
• Pro filter — score ≥6/8 · 9:45 AM–3:45 PM ET (through force-flat)

Requirements (local install only — the hosted connector above needs none)

1. Python 3.10+
2. pip install "mcp<2" tzdata — pin below 2.0 (mcp 2.x removed FastMCP and breaks the server); tzdata is required on Windows
3. Pro API key from success-pro (after checkout)
4. Path to scripts/confluence_mcp/server.py in your repo clone

Cursor

Add to .cursor/mcp.json (adjust path):

{
  "mcpServers": {
    "confluence-pro": {
      "command": "python3",
      "args": ["/path/to/Agent/scripts/confluence_mcp/server.py"],
      "env": {
        "CONFLUENCE_API_KEY": "cfl_your_key",
        "CONFLUENCE_API_BASE": "https://api.0dteconfluence.com"
      }
    },
    "robinhood-trading": {
      "url": "https://agent.robinhood.com/mcp/trading"
    }
  }
}

Restart Cursor → Settings → MCP → enable both servers.

Claude Desktop

Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "confluence-pro": {
      "command": "python3",
      "args": ["/path/to/Agent/scripts/confluence_mcp/server.py"],
      "env": {
        "CONFLUENCE_API_KEY": "cfl_your_key",
        "CONFLUENCE_API_BASE": "https://api.0dteconfluence.com"
      }
    }
  }
}

Restart Claude Desktop. Add Robinhood MCP separately if your Claude build supports remote MCP URLs.

Windows: use "command": "python" (not python3) and forward slashes in the path (e.g. C:/Users/you/confluence_mcp/server.py). Install Claude from claude.ai/download — the Microsoft Store build is sandboxed and often can’t launch local MCP servers. After editing the config, fully quit Claude (system tray → Quit) and reopen.

Windows config block:

{
  "mcpServers": {
    "confluence-pro": {
      "command": "python",
      "args": ["C:/Users/you/confluence_mcp/server.py"],
      "env": {
        "CONFLUENCE_API_KEY": "cfl_your_key",
        "CONFLUENCE_API_BASE": "https://api.0dteconfluence.com"
      }
    }
  }
}

VS Code

Install an MCP extension for VS Code, then point it at the same command, args, and env as Cursor (Confluence Pro block above). Exact config file depends on the extension — use your checkout success page JSON as the template.

Session workflow

1. get_morning_desk + get_rails before the open
2. Swing / weekly: get_swing_structure (5–45 DTE walls — not 0DTE)
3. Macro overlay: get_cross_asset · get_major_news · get_yield_curve · get_fear_greed (context only)
4. When Discord posts: get_latest_alert (same card as #alerts-a / #alerts-b) → check_alert_tradeable → plan_robinhood_entry
5. Robinhood MCP (or app): chain → quotes → review → you confirm → order
6. After fill: format_bracket_plan with your entry premium

Tradeability gate + fill-rate pricing

get_latest_alert and list_alert_history return only setups Discord actually posted — same windows, quality gates, and card copy as #alerts-a / #alerts-b (plain-text card field). Scanner grades Discord skipped (lunch B, tape conflict, rail cooldown) do not appear in MCP.

check_alert_tradeable is the hard gate before you size anything. It fails closed when the alert is stale, outside the Pro window (9:45–force-flat), score < 6/8, or missing level/direction. Only a tradeable: true result should reach Robinhood.

plan_robinhood_entry then returns the advisory checklist — ~0.40–0.55 delta, spread ≤10% of mid (≤12% pre-9:45), and entry priced from Robinhood’s high_fill_rate_buy_price (not a guessed ask or passive mid). Exits use low_fill_rate_sell_price when available. Bracket defaults: scale toward structure, −25% premium cap, force-flat from get_market_clock.

1. check_alert_tradeable  → tradeable must be true
2. get_option_quotes      → high_fill_rate_buy_price = your buy limit
3. review_option_order    → fees / collateral / order_checks
4. place_option_order     → you confirm
5. format_bracket_plan    → +25% / −25% (or structural) on filled qty

Discord backup

#alerts-pro pings Pro members on mobile. MCP is the primary desk workflow when you're at the computer.

Install guide above · questions: [email protected]