Screen options, analyze portfolios, and compare strategies — all from a conversation with your AI.
Wherever you connect, this is the address you paste. You sign in with your existing Remora login — the AI is never handed your password or an API key.
https://mcp.heyremora.com/mcp
Open Claude's connector settings →
If you logged into Claude Code with your Claude account, a connector you added on the web is available automatically — run /mcp to see it. (This does not apply when ANTHROPIC_API_KEY, Bedrock/Vertex, or a setup-token is the active auth — run /status, then /login.)
claude mcp add --transport http --scope user remora https://mcp.heyremora.com/mcp
Then run /mcp, select remora, and choose Authenticate — a browser opens for the Remora sign-in. --scope user makes it available in every project.
If a long screen ever trips the 60-second first-response timer, raise the per-server timeout:
claude mcp add-json --scope user remora '{"type":"http","url":"https://mcp.heyremora.com/mcp","timeout":120000}'
ChatGPT connects to Remora as a custom connector. It cannot run the local (pip) server, so this is the path — the same connector URL as Claude.
/mcp path — the connector URL above.Cursor and VS Code both speak the remote server, so neither needs the local package or an API key. The local (pip) path is the next tab if you would rather run the server yourself.
One click: Add to Cursor. Cursor asks to install the server, then authenticates with your Remora login.
Or add it by hand, in ~/.cursor/mcp.json (or .cursor/mcp.json for one project) — a remote server is the entry with a url:
{
"mcpServers": {
"remora": {
"url": "https://mcp.heyremora.com/mcp"
}
}
}
Then open a chat and ask for something. If Cursor asks you to sign in, sign in with your Remora login and approve. Installed servers are managed on Cursor's Customize page.
Add the remote server to mcp.json. MCP: Open User Configuration opens the one in your user profile; .vscode/mcp.json is the per-workspace file.
{
"servers": {
"remora": {
"type": "http",
"url": "https://mcp.heyremora.com/mcp"
}
}
}
Or from the command line — MCP: Add Server in the Command Palette is the guided version of the same thing:
code --add-mcp "{\"name\":\"remora\",\"type\":\"http\",\"url\":\"https://mcp.heyremora.com/mcp\"}"
VS Code shows a dialog the first time the server starts: review it, then sign in with your Remora login when the browser opens. Configure Tools in the chat input lists Remora's tools.
url entry from cursor.com/docs/mcp; the VS Code servers / type: http entry and --add-mcp from code.visualstudio.com/docs/copilot/customization/mcp-servers.
For offline use, or any client that runs a local (stdio) MCP server. This path uses a personal API key and works today. Cursor and VS Code can use it too, though both reach the remote server directly.
Go to your Account page → API Keys, click Create Key, name it (e.g. "My AI"), and copy it.
pip install heyremora
Requires Python 3.10+. This installs the heyremora command.
Open your client's MCP config file. For Claude Desktop (macOS):
~/Library/Application Support/Claude/claude_desktop_config.json
Add the following, replacing the placeholder with your key:
{
"mcpServers": {
"remora": {
"command": "heyremora",
"env": {
"REMORA_API_KEY": "otx_live_YOUR_KEY_HERE",
"REMORA_API_URL": "https://heyremora.com"
}
}
}
}
Quit and reopen it. You should see Remora in the tools list. Try:
Find me cash-secured puts on AAPL and NVDA with delta between 0.20-0.35 and at least 30 DTE
Check that your REMORA_API_KEY in the config matches exactly what was shown when you created the key. Keys start with otx_live_.
Verify REMORA_API_URL is set correctly. For the production API, use https://heyremora.com.
API keys need an active trial or a paid plan. If your trial has expired, existing keys stop authenticating until you subscribe — the keys themselves are not deleted. Check your status on the Account page.
1. Make sure the package is installed: pip install heyremora
2. Check the config file path is correct (Claude Desktop, macOS: ~/Library/Application Support/Claude/claude_desktop_config.json)
3. Restart your AI assistant completely (Quit, not just close the window)
4. Check MCP logs for error details (Claude Desktop, macOS: ~/Library/Logs/Claude/)
Pro accounts have generous rate limits (100 req/min). If you hit the limit, wait a minute and try again. Avoid running bulk screens in tight loops.
See the Tool Reference for the full list of tools (45+) and their parameters.