Early preview — building the MCP layer for title

Composable title automation,
spoken in MCP.

TitleMCP is an open-source collection of MCP servers, tools, and jurisdiction-aware adapters for the real estate title industry — reusable primitives for estoppels, municipal liens, payoffs, OCR, and curative workflows.

$pip install titlemcp
server.py
python
from titlemcp.server import build_server, tool
from titlemcp.schemas import AssessmentRecord
from .auditor import FranklinCountyAuditor

server = build_server(name="titlemcp-us-oh-franklin-recorder")

@tool(name="franklin_county_auditor_search", output_schema=AssessmentRecord)
async def search(parcel_id: str | None = None, address: str | None = None):
    """Look up the Franklin County, OH auditor record for a parcel."""
    raw = await FranklinCountyAuditor().lookup(parcel_id=parcel_id, address=address)
    return AssessmentRecord.from_franklin(raw)

server.run(transport="stdio")

Built on open standards

Model Context ProtocolPython 3.12+PydanticAnthropicOllamaAWS TextractOpenAPI

How it actually works

You ask. The model picks the tool.
The tool does the work.

TitleMCP turns title operations into conversational tools any MCP-aware LLM can call — Claude, GPT, or a local Ollama model — with audited, schema-typed results.

You

I'm reviewing a Franklin County, Ohio property. Can you look up the county auditor record for parcel 030-000526-00 and return the assessment record?

Model decides to call an MCP tool:

→ tool franklin_county_auditor_search { parcel_id: "030-000526-00" }

Returns a typed AssessmentRecord — auditable, no hallucinated fields.

The problem

Title operations run on brittle scripts and PDFs.

Every title shop rebuilds the same workflows from scratch — HOA estoppels, municipal lien searches, tax certs, payoff parsing, document classification. The integrations are jurisdictional, the formats are inconsistent, and the tooling is locked inside vendor portals.

Meanwhile, AI agents are emerging that could reason across these workflows — if only there were a stable, structured interface to plug into.

TitleMCP is that interface. A shared, open protocol layer for title.

50+
Counties with bespoke estoppel formats
3,143
U.S. counties — no shared schema
~70%
of curative work is still manual

Primer

What is MCP?

The Model Context Protocol is the connective tissue between LLMs and the real systems title work depends on — county auditors, payoff portals, HOA management software, document stores, and more.

A shared language

MCP (Model Context Protocol) is an open standard from Anthropic that lets AI assistants talk to external tools and data sources through one consistent interface.

Tools the model can call

Each MCP server exposes typed tools — like 'lookup_parcel' or 'order_estoppel'. The model decides when to call them, and the server does the actual work.

Composable by default

Any MCP-aware client (Claude, Cursor, Ollama, your own agent) can connect to any MCP server. Build once, use anywhere — no bespoke integrations.

The future of title
should be open.

Star the repo, ship an adapter, or wire your first MCP-powered title workflow today.