Tickerbot in your AI tool.

Install the MCP server once and Claude, ChatGPT, Cursor, or any other Model Context Protocol client can scan, fetch, and subscribe in plain English. Six install paths below, ordered from easiest to hardest. Live in ChatGPT, your IDE, and terminal — Claude.ai and Cursor marketplace listings landing soon.

what you’ll ask it

Every endpoint, addressable in chat.

Once installed, the SQL grammar Tickerbot speaks becomes something you describe in English. A few prompts that land out of the box:

tier 1 · marketplace

One-click from inside your chat app or IDE. Coming soon

The lowest-friction path. Open your chat app or IDE’s built-in connector library, find Tickerbot, click install. No URLs, no config files, no terminal.

ChatGPT · GPT Store

Open in ChatGPT →
Listed in ChatGPT's GPT Store as “Tickerbot — the stock market, in SQL.” Open the link, paste a tb_live_… key from your dashboard when prompted, and ask the live market in plain English.

Claude.ai · Connectors library

Pending Anthropic review

Listed alongside Stripe, Linear, Notion in claude.ai's Connectors menu. Sign in to Tickerbot via OAuth and Claude gets the keys — no URL paste, no config file.

Blocked by /mcp OAuth + Anthropic vetting.

Cursor · MCP catalog

Pending Cursor registry

One-click install from Cursor's built-in MCP browser. Cursor consults the public MCP Registry where Tickerbot is already listed — surfaces in their UI as their catalog sync runs.

Tickerbot is on the public MCP Registry; awaiting Cursor sync.

tier 2 · desktop extension

Download an installer, double-click. Coming soon

Anthropic’s Desktop Extensions (.dxt) are the install-button experience for Claude Desktop — native install dialog, UI prompt for your API key, no config-file editing.

Claude Desktop · Desktop Extension (.mcpb)

Download tickerbot.mcpb →
Two install paths, either works: double-click the file after download, or open Claude Desktop → Settings → Extensionsdrag the file into the panel. Claude shows a native install dialog with a “Tickerbot API key” field — paste a tb_live_… key from your dashboard and restart Claude.

Submitting to Anthropic's Extensions catalog for one-click in-app install (separate, long-running approval).

tier 3 · chat-app custom connector

Paste a URL and your API key into chat-app settings. Coming soon

For Claude.ai or ChatGPT power users on Pro / Team plans who don’t want to wait for marketplace approval. Paste https://api.tickerbot.io/mcp in your chat app’s custom-connector settings.

Claude.ai (web + mobile) · Custom Connector

Blocked on OAuth

Settings → Connectors → Add custom connector. Paste https://api.tickerbot.io/mcp. Today Claude.ai's UI only accepts OAuth credentials — we're wiring an OAuth flow on our endpoint to unblock this path for power users.

OAuth on /mcp in progress.

ChatGPT (Pro / Team) · Custom Connector

Blocked on OAuth

Settings → Connectors → Add connector. Same OAuth requirement as Claude.ai — same fix unlocks both.

OAuth on /mcp in progress.

tier 4 · CLI command

One line in your terminal.

For terminal-native AI coding agents. Each tool exposes an mcp add command that wires Tickerbot in one step, no JSON editing.

Claude Code

One-line install. The CLI handles the config wiring and restart for you.

Run in your terminal
claude mcp add tickerbot --env TICKERBOT_API_KEY=tb_live_... -- npx -y @tickerbot/mcp-server

Codex CLI · Aider · OpenClaw · other MCP-aware CLIs

Each tool exposes its own mcp add equivalent that takes the same shape: an env var for TICKERBOT_API_KEY and an npx command running @tickerbot/mcp-server. Refer to each tool's MCP docs for the exact syntax.

tier 5 · JSON config

Edit a config file, paste the snippet, restart.

Universal install path for every MCP-aware desktop app and IDE. Same npx command everywhere — only the config file location changes.

Claude Desktop

Edit the desktop app's config file, paste the block below, restart Claude Desktop.

Config file
~/Library/Application Support/Claude/claude_desktop_config.json (macOS)
%APPDATA%\Claude\claude_desktop_config.json (Windows)
{
  "mcpServers": {
    "tickerbot": {
      "command": "npx",
      "args": ["-y", "@tickerbot/mcp-server"],
      "env": { "TICKERBOT_API_KEY": "tb_live_..." }
    }
  }
}

Once the DXT bundle ships (Tier 2), this manual path becomes the fallback for users who prefer to edit config.

Cursor

Paste into Cursor's MCP config file, reload the window.

Config file
~/.cursor/mcp.json
{
  "mcpServers": {
    "tickerbot": {
      "command": "npx",
      "args": ["-y", "@tickerbot/mcp-server"],
      "env": { "TICKERBOT_API_KEY": "tb_live_..." }
    }
  }
}

VS Code (Copilot)

Paste into the workspace or user MCP config file.

Config file
.vscode/mcp.json (workspace) or ~/.config/Code/User/mcp.json (user)
{
  "servers": {
    "tickerbot": {
      "command": "npx",
      "args": ["-y", "@tickerbot/mcp-server"],
      "env": { "TICKERBOT_API_KEY": "tb_live_..." }
    }
  }
}

Windsurf

Paste into Windsurf's MCP config.

Config file
~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "tickerbot": {
      "command": "npx",
      "args": ["-y", "@tickerbot/mcp-server"],
      "env": { "TICKERBOT_API_KEY": "tb_live_..." }
    }
  }
}

Zed

Paste into Zed's settings under the context-servers section.

Config file
~/.config/zed/settings.json (under "context_servers")
{
  "context_servers": {
    "tickerbot": {
      "command": {
        "path": "npx",
        "args": ["-y", "@tickerbot/mcp-server"],
        "env": { "TICKERBOT_API_KEY": "tb_live_..." }
      }
    }
  }
}

tier 6 · BYO runtime

Writing your own agent loop?

If you’re building with LangChain, the Vercel AI SDK, the OpenAI SDK, the Anthropic SDK, or raw HTTP, you can skip MCP entirely. Drop the Tickerbot tool definitions into your runtime’s tool-use schema.

See /api/agents for the raw JSON: Claude tool-use shape, OpenAI function-calling shape, and MCP tools/list shape. Same tools, three formats. Pick the one your runtime expects.

polish layer · skills, gems, agents

Make the LLM smarter at using Tickerbot. Coming soon

Not an install path — these ride on top of the tool surface. Bundled instructions that teach the LLM the q-grammar, schema, and scan/replay/subscribe playbook so it picks the right tool every time. One playbook, different distribution per vendor.

Claude Skill

Building

A tickerbot-analyst Skill that bundles the q-grammar, full schema, and scan/replay/subscribe playbook. Loads into Claude's context across Desktop, Code, and web. Works on top of the MCP tools you already installed.

Gemini Gem

Building

Gemini doesn't speak MCP — a Tickerbot Gem would call the REST API directly via Gemini's function-calling layer. Same brand voice, same tool surface, different wire.

Microsoft Copilot agent

Building

Copilot Studio agent that wraps the REST API for users inside the Microsoft 365 ecosystem.

the tool surface

19 tools cover every public endpoint.

Strategies are intentionally not exposed via MCP — see the REST API docs for those.

try it

Get a key and install it in five minutes.