MCP connector at mcp.tipranks.com · Production

TipRanks financial data - built for AI agents

Plug Claude, ChatGPT, Cursor, or any MCP client into live analyst ratings, Smart Score, technicals, earnings, and your TipRanks portfolios. One URL, OAuth or API key, ready in 60 seconds.

Same TipRanks data that powers research at 100+ banks, brokers, and exchanges - including Robinhood, E*TRADE, Interactive Brokers, Santander, and TMX - plus tipranks.com (10M+ monthly users).

curl -X POST https://mcp.tipranks.com/dev/api/signup \
  -H 'content-type: application/json' \
  -d '{"email":"agent@example.com","password":"<10+ chars>","label":"my-agent"}'

What you can do

See it in action

Real screenshots of Claude.ai using this connector. Each one is a single prompt - the chart, dashboard, or table is rendered inline by Claude's Analysis tool from JSON returned by TipRanks MCP tools.

See the full playbook catalog ↓

Quickstart

Two ways to authenticate. Pick whichever fits the client.

Remote - OAuth (recommended for Claude.ai & Claude Desktop)

https://mcp.tipranks.com/mcp/

Sign in with your TipRanks account when the client prompts. Grants tipranks:portfolio if you opt in.

Remote - API key (recommended for ChatGPT, Cursor, Codex, agents)

https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY

Get a free key at /dev/signup or via the JSON API (see Run locally).

Claude.ai (web & desktop)

The full OAuth flow - sign in with TipRanks, approve scopes, done. Claude.ai handles token refresh automatically.

1. Open Claude.ai → Settings → Connectors (or the connector panel in Claude Desktop).
2. Click Add custom connector.
3. Enter the URL: https://mcp.tipranks.com/mcp/
4. Click Connect. A new tab opens to TipRanks for sign-in (skipped if you already have a TipRanks session).
5. On the consent screen, approve tipranks:read (market data) and optionally tipranks:portfolio (your TipRanks portfolios).

You'll need a TipRanks account. Free, Premium, and Ultimate plans all work; per-account monthly quotas differ - see Scopes & quotas.

Claude Desktop config file

If you prefer editing the config file directly (or you're on an older Claude Desktop build that only supports stdio), point at the hosted server through the mcp-remote proxy with your API key:

{
  "mcpServers": {
    "tipranks": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY"
      ]
    }
  }
}

Replace YOUR_API_KEY with the key shown after signup. Save the file at:

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

Restart Claude Desktop and the tipranks tools appear in the 🎲 menu.

Claude Code CLI

One command - works against the hosted server. Replace YOUR_API_KEY with a real key.

claude mcp add -t http tipranks https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY

Or, if you'd rather use OAuth (and don't mind the consent screen popping up in your browser):

claude mcp add -t http tipranks https://mcp.tipranks.com/mcp/

List or remove later with claude mcp list / claude mcp remove tipranks.

ChatGPT (developer mode)

Available on ChatGPT Plus and above. The MCP server is registered as a custom app.

1. Open ChatGPT → Settings → Apps.
2. Click Advanced settings → enable Developer mode.
3. Return to Apps, click Create app.
4. MCP Server URL: https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY
5. Authentication: No authentication (the key is in the URL).

Cursor

Edit ~/.cursor/mcp.json (user-wide) or .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "tipranks": {
      "url": "https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY"
    }
  }
}

Reload Cursor and the tools appear under the MCP indicator.

VS Code

Add an HTTP MCP server in .vscode/mcp.json (workspace) or the user-level equivalent:

{
  "servers": {
    "tipranks": {
      "type": "http",
      "url": "https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY"
    }
  }
}

OpenAI Codex

Codex CLI reads ~/.codex/config.toml. Use the mcp-remote proxy because Codex's MCP transport is stdio-only:

[mcp_servers.tipranks]
command = "npx"
args = ["-y", "mcp-remote", "https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY"]

Gemini CLI

One-shot command:

gemini mcp add -t http tipranks https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY

Or edit ~/.gemini/settings.json directly:

{
  "mcpServers": {
    "tipranks": {
      "httpUrl": "https://mcp.tipranks.com/mcp/?apikey=YOUR_API_KEY"
    }
  }
}

Run locally / programmatic signup

Programmatic signup (one HTTP call, no email verification)

Agents and CI jobs can mint a free-tier key in one round-trip:

POST https://mcp.tipranks.com/dev/api/signup
Content-Type: application/json

{ "email": "you@example.com", "password": "Aa1!aaaaaaaa", "label": "my-agent" }

Returns 201 with a usable key:

{
  "account_id": "dev_...",
  "api_key": "tr_live_...",
  "tier": "free",
  "rpm_limit": 5,
  "daily_limit": 25,
  "manage_url": "https://mcp.tipranks.com/dev/login",
  "mcp_url": "https://mcp.tipranks.com/mcp"
}

Self-host from source

The full server is open source. Clone the repo and run it in stdio mode for any MCP client that spawns local processes:

git clone https://github.com/tipranks/tipranks-python-apps
cd tipranks-python-apps
pip install -r mcp_server/requirements.txt

# Stdio mode (for Claude Desktop / Cursor / Codex local config):
cd mcp_server
MCP_TRANSPORT=stdio python -m server

# Or HTTP mode (mirrors the hosted deployment):
MCP_TRANSPORT=http python -m server

Stdio mode bypasses the OAuth and API-key middleware - it's intended for a single trusted user on the local machine, not a public server. For a public deployment, use the HTTP transport with the OAuth / API-key auth stack.

Scopes & quotas

OAuth users (Claude.ai connector)

Tool calls are counted per calendar month per TipRanks user. Quota resets on the 1st of every month (UTC).

API key users

Per-key rate limits. Every tier carries both a per-minute and a per-day cap.

Hitting any cap returns HTTP 429 with X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Window, and Retry-After headers so clients can back off intelligently. Upgrade plans from the developer portal; the new limits cascade onto every active API key on your account.

Example prompts

Once connected, ask the model questions like these. The tool calls happen automatically.

Playbooks

The examples above are single questions. The connector really shines on multi-step research workflows - Claude calls one tool, looks at the result, calls another, then synthesizes. These are copy-pasteable starting points; tweak tickers / sectors / time windows to match what you're after.

Charts & visualizations

You don't need anything special on the connector side to get charts - that part comes from the client.

On Claude.ai and Claude Desktop, the built-in Analysis tool (a sandboxed JS/Python runtime) can take any time-series data an MCP tool returns and render it as a chart in the conversation. Just ask - e.g. “chart NVDA's daily close for the past year with 50- and 200-day moving averages” - and Claude will call the relevant tool (here get_assets_data or one of the historical tools), pipe the JSON through the Analysis tool, and produce a plot inline. The same works for technical indicators, multi-asset comparisons, and portfolio performance breakdowns.

Tools that produce chart-friendly output include the historical OHLCV endpoints (get_commodity_historical, get_forex_historical, get_crypto_historical, get_index_historical), the per-ticker financial statements time series (get_financials), the per-ticker earnings and dividend time series (get_earnings_history, get_dividend_history), the per-ticker buyback series (get_buybacks with a ticker argument), and the portfolio performance tools (get_portfolio_performance, get_portfolio_overview).

Tools reference

Every tool is read-only, idempotent, and returns JSON. Categories follow the structure of server.py.

Asset data

Discovery & screening

Alternative data & signals

Signals beyond price and ratings - what political / insider / hedge-fund money is doing, where the biotech catalysts are, who's buying back stock.

Macro & calendars

Technical analysis

ETFs

Commodities

Forex

Indices

Crypto

Portfolios requires tipranks:portfolio

Privacy & support

The full connector-specific privacy policy is at /privacy. It supplements the broader TipRanks privacy policy which governs your TipRanks account itself.

In short: we store hashed OAuth tokens, your TipRanks user ID / email / tier, the portfolio IDs you authorized, and a per-user monthly request counter. We do not store tool-call payloads, the financial data we return, or your TipRanks credentials - sign-in happens on tipranks.com, never here.

Authentication: standard OAuth 2.1 with Dynamic Client Registration (RFC 7591) and PKCE-S256. Public clients only (token_endpoint_auth_method=none). Discovery at:

GET https://mcp.tipranks.com/.well-known/oauth-authorization-server

Tokens are SHA-256 hashed at rest; refresh tokens rotate on use, with chain-revocation on replay.

• Connection or auth issues: open a ticket in your TipRanks account.
• API behavior bugs or feature requests: support@tipranks.com.
• Privacy / data subject requests: see the privacy policy.

Last updated: 2026-05-19 · Hosted at https://mcp.tipranks.com/docs