Python · Agent framework

Use OpenRegistry from LlamaIndex

Use OpenRegistry from LlamaIndex agents and workflows via the official MCP tool spec.

LlamaIndex's BasicMCPClient connects to any MCP server. Pass the resulting McpToolSpec to any LlamaIndex agent — ReActAgent, FunctionAgent, AgentWorkflow — and your model can call any OpenRegistry tool directly.

Install

pip install llama-index-tools-mcp llama-index-llms-anthropic

Minimum working example

from llama_index.tools.mcp import BasicMCPClient, McpToolSpec
from llama_index.core.agent.workflow import ReActAgent
from llama_index.llms.anthropic import Anthropic

mcp = BasicMCPClient(
    "https://openregistry.sophymarine.com/mcp",
    transport="streamable_http",
)
tools = await McpToolSpec(mcp).to_tool_list_async()

agent = ReActAgent(tools=tools, llm=Anthropic(model="claude-3-5-sonnet-latest"))
resp = await agent.run(
    "Walk the ownership chain of Revolut Ltd (GB) until you reach a natural person."
)
print(resp)

What you get

27 tools — search, profile, officers, PSC/UBO, shareholders, filings, XBRL accounts, charges, documents. See the full catalogue.
10 named workflow prompts — KYC dossier, UBO chain-walker, shell-company detector, phoenix-fraud radar. See all 10.
27 national registries — UK Companies House, Norway Brreg, Germany Handelsregister, France RNE, Spain BORME, Korea OpenDART, Canada CBCA, and more. See the map.
Live + source-linked — every field traces back to the government register. No aggregator markup.
Free tier: 20 req/min per IP, no signup, no API key.

Framework docs

Official MCP integration docs for LlamaIndex:

docs.llamaindex.ai — MCP tools ›

Use from a different framework?

OpenRegistry is a Model Context Protocol server. Any MCP client can connect — these are just the most popular agent frameworks.