No precise location inputs
OpenAI says to rely on the coarse location the system shares and not to request precise user location, so latitude, longitude or GPS arguments block review; addresses are flagged unless they are clearly task data.
Check openai-precise-location · OpenAI plugin directory · default severity: blocking (or attention, depending on the case) · rubric 2.1.0
What fails
Coordinate fields (latitude, longitude, gps) carry a precise location and are blocking. Street address, postal code and location fields are marked for attention: a user's own address is a precise location, but a field that is not the user's location (a shipping destination, a store) may be fine if the description says so. City-level fields are coarse and are not flagged.
The rule
When location is needed, rely on system-shared coarse location. Do not request precise user location.
How to fix it
Use the coarse location the platform shares. If an address is task data, not the user's location, say that in the field's description.
- Remove latitude, longitude and gps arguments that capture the user’s position.
- Use city or region-level input, or the coarse location the platform provides.
- If an address is task data (a delivery destination, a store), say so in the field description.
TypeScript SDK
import { McpServer } from '@modelcontextprotocol/server';
import * as z from 'zod';
const server = new McpServer({ name: 'issues', version: '1.0.0' });
server.registerTool(
'find_stores',
{
title: 'Find stores',
description: 'List stores in a city. Use when the user asks where to buy something nearby.',
inputSchema: z.strictObject({
city: z.string().describe('City name, for example Tucson'),
limit: z.number().int().min(1).max(20).default(5).describe('Maximum stores, 1 to 20'),
}),
annotations: { readOnlyHint: true, destructiveHint: false, openWorldHint: false },
},
async ({ city, limit }) => ({ content: [{ type: 'text', text: 'Done' }] }),
);Python SDK
from typing import Annotated
from pydantic import Field
from mcp.server.mcpserver import MCPServer
from mcp.types import ToolAnnotations
server = MCPServer("issues")
@server.tool(
title="Find stores",
annotations=ToolAnnotations(readOnlyHint=True, destructiveHint=False, openWorldHint=False),
)
def find_stores(
city: Annotated[str, Field(description="City name, for example Tucson")],
limit: Annotated[int, Field(description="Maximum stores, 1 to 20", ge=1, le=20)] = 5,
) -> str:
"""List stores in a city. Use when the user asks where to buy something nearby."""
return "Done"Run against @modelcontextprotocol/server 2.4.0, mcp 2.3.0 and fastmcp 4.2.2 on 2026-11-18; the checker confirms this check passes on the tool list each snippet produces.
Questions
Can a delivery app take a shipping address?
Yes, when it is the destination of an order, which is task data. Describe the field as the delivery address so a reviewer sees it is not the user’s location.
Is a postal code precise?
It can identify a small area, so the checker marks it for attention. City-level fields are not flagged.
Related checks
- No conversation history or "just in case" context inputs
- No credential, payment card or government ID inputs
- Every argument has a description
- All checks in Directory review rules for MCP tools
How this check is graded
It counts toward: OpenAI plugin directory. See the rubric.
Last verified against OpenAI plugin guidelines, section 6.4 on 2026-11-18. The rule above is quoted in the source’s own words. This page as Markdown.
Check your tool list: paste your tools/list, nothing is sent anywhere