Blocksize Remote MCP Quickstart
Public remote discovery endpoint: https://mcp.blocksize.info/mcp/server/
The public remote MCP exposes free discovery and documentation tools: instrument search,
supported instrument lists, pricing inspection, and document-style search/fetch.
Search AAPL with asset_class=equity to discover supported stock tickers,
then route live equity snapshots through /v1/bidask/{ticker}.
Live paid market data remains available through the x402-protected HTTP API and the advanced local MCP setup.
ChatGPT Developer Mode
OpenAI’s current docs say developer mode supports remote MCP servers over Streamable HTTP. In ChatGPT, enable Developer mode under Settings, then create or add an app pointing at the remote MCP URL above.
Reference docs: Developer mode and remote MCP overview.
Cursor
Cursor’s MCP docs say Streamable HTTP servers can be configured with a URL. Add this to
.cursor/mcp.json or ~/.cursor/mcp.json:
{
"mcpServers": {
"blocksize-discovery": {
"url": "https://mcp.blocksize.info/mcp/server/"
}
}
}
In Cursor, ask: Search Blocksize for AAPL with asset_class=equity and build the paid bid/ask URL.
Anthropic Messages API
Anthropic’s MCP connector docs say remote servers must be publicly exposed over HTTP and support
Streamable HTTP or SSE. Use the public MCP URL in the mcp_servers array:
{
"mcp_servers": [
{
"type": "url",
"url": "https://mcp.blocksize.info/mcp/server/",
"name": "blocksize-discovery"
}
]
}
Claude Custom Connector
For authenticated read-only live market data in Claude, use the dedicated Claude connector: https://mcp.blocksize.info/claude-connector. That connector uses OAuth and server-side daily data credits, and does not execute wallet transactions or submit payment proofs.
In Claude, ask: Search Blocksize equities for AAPL, then fetch the AAPL bid/ask snapshot if it is supported.
Any Streamable HTTP Client
If your agent runtime supports remote MCP over Streamable HTTP, point it at the same URL:
https://mcp.blocksize.info/mcp/server/.
Paid Data Access
For live paid market data, use the REST API documented in Swagger and OpenAPI JSON. The public pricing and payment model are documented in the Pricing Guide and Agent Manual.
Advanced Local MCP
Approved collaborators who want the full MCP tool surface can run the private local server:
uv venv
source .venv/bin/activate
uv pip install -e ".[dev]"
python -m src.mcp_server
The local MCP flow requires Blocksize credentials and is intended for direct builder integrations, not anonymous public directory installs.