lease-qa-mcp
This server exposes a single MCP tool, lease_qa, for asking natural language questions about lease terms and receiving grounded answers from a LeaseClear instance. It runs as a local stdio process, keeping lease data in your own environment. You can optionally scope queries to specific leases via document_ids, and authenticate with a LEASECLEAR_API_KEY to query your own leases or omit it to use the public demo corpus. Configure LEASECLEAR_BASE_URL to point to your LeaseClear instance. The tool returns structured responses like { "answer": "..." }, including explicit acknowledgment when the lease is silent. It integrates with MCP hosts such as Cursor and Claude Desktop, and errors are surfaced as MCP tool errors.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@lease-qa-mcp Does this lease allow subletting?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
leaseclear-mcp
A Model Context Protocol server that exposes LeaseClear lease Q&A.
Listed in the official MCP Registry as io.github.amadeuserras/leaseclear-mcp.
Tools
lease_qa— ask one question about lease terms; returns an answer grounded in the lease, or states that the lease is silent
Related MCP server: MCP RAG Server
Usage
Add this to your MCP client config (e.g. claude_desktop_config.json or Cursor mcp.json):
{
"mcpServers": {
"lease-qa": {
"command": "uvx",
"args": ["leaseclear-mcp"]
}
}
}The package is on PyPI. uvx fetches and runs it as a local subprocess, communicating over stdio. uv required.
Optional environment variables
Variable | Description | Default |
| Use the server with your own LeaseClear account | LeaseClear demo mode |
| Override the API endpoint (local or private instances) | LeaseClear production backend |
{
"mcpServers": {
"lease-qa": {
"command": "uvx",
"args": ["leaseclear-mcp"],
"env": {
"LEASECLEAR_API_KEY": "lc_...",
"LEASECLEAR_API_URL": "https://..."
}
}
}
}lease_qa usage
Argument
question(string, required) — one question about the lease
_meta
document_ids(string[], optional) — UUIDs of the documents to query. Omitted means all.
Security: The _meta field is used to pass metadata to the tool call that the model doesn't see. Having document_ids there is intentional: it prevents cross-document prompt injection because the model is never able to choose which documents it has access to. See a real-world example in LeaseOps: prompt injection and the tenants table.
MCP Python SDK example
Tool call:
result = await session.call_tool(
name="lease_qa",
arguments={
"question": "How much is the security deposit for Yuna Kim?"
},
meta={
"document_ids": [
"a1b2c3d4-e5f..."
]
}
)Output:
{
"answer": "The security deposit is $6,400.00. This deposit is held in Owner's Broker's trust account [california-johnson-kim §4]."
}Tech stack
Python 3.12
Model Context Protocol Python SDK (
mcp[cli]) over stdiohttpxfor LeaseClear API callsPydantic and
pydantic-settingsfor schemas and configuvanduv_buildfor running and packagingpytest, Ruff, and Pyright for tests, linting, and type checking
Project structure
leaseclear-mcp/
├── src/leaseclear_mcp/
│ ├── server.py # MCP server + lease_qa
│ ├── leaseclear.py # LeaseClear HTTP client
│ ├── schemas.py
│ └── config.py
├── tests/
├── pyproject.toml
├── .env.example
└── README.mdLocal Development
git clone https://github.com/you/leaseclear-mcp.git
cd leaseclear-mcp
uv sync
cp .env.example .env
uv run pytest
uv run ruff check .
uv run pyrightDebugging
Use the MCP Inspector to test the server interactively:
npx @modelcontextprotocol/inspector uvx leaseclear-mcpFrom a checkout, use uv run leaseclear-mcp instead of uvx leaseclear-mcp.
License
MIT
Available Tools
1 toollease_qaA
Ask one neutral, precise question about the tenant's lease. The lease document is already scoped for this email. Returns an answer grounded in the lease, or states that the lease does not address the question.
Inputs: question (required). Returns: answer
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 1 tool
With only one tool, there is no possibility of confusion or overlap. The tool's purpose is clear and unique.
A single tool name cannot be inconsistent with others. The name 'lease_qa' is straightforward and matches the server's purpose.
Having only one tool feels thin for a typical server, but it may be appropriate for this narrow scope. The calibration suggests 1-2 tools is borderline.
The tool directly fulfills the stated purpose of asking lease questions and getting grounded answers. Minor gaps exist, such as no ability to retrieve the full lease or ask multiple questions in a batch, but the core workflow is covered.
Related MCP Connectors
An MCP server that provides an API to LLMs to manage their JumpCloud resources.
MCP server for querying Forkast documentation
MCP server providing access to the Scorecard API to evaluate and optimize LLM systems.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables querying, extracting terms, and projecting rent from commercial lease documents via MCP tools.-
- FlicenseNot gradedqualityDmaintenanceMCP server for retrieving answers from local PDFs using RAG with FAISS and OpenAI.-
- FlicenseNot gradedqualityDmaintenanceAn MCP server that exposes grounded, source-attributed question-answering over a collection of PDF documents.-
- AlicenseNot gradedqualityDmaintenanceMCP server for grounded agentic Q&A over customer feedback, exposing typed tools to query a feedback corpus and return answers with citations to specific record IDs or a refusal when unsupported.MIT