pointmoon-mcp
Provides sourced, current physical and environmental field-truth (weather, air quality, water, terrain, etc.) for any location, with provenance and typed silence to prevent hallucination.
README
pointmoon-mcp
Ground truth for agents acting in the physical world. Sourced, time-stamped, or silent.
Pointmoon is an agent-callable trust layer. Give it a coordinate (or a place name) and it returns sourced, current physical and environmental field-truth: weather, air quality, light and sky, water, terrain, notable natural events. Every fact it returns carries its own provenance — which provider observed it (source), when (observedAt), how long it stays fresh (ttlMinutes), and a confidence score. When a fact is unknown, stale, or low-confidence, Pointmoon returns typed silence with a reason instead of inventing a value.
That is the whole point: a fluent model is good at language and bad at knowing whether it is making the weather up. Pointmoon refuses to hallucinate about physical reality. It hands your model sourced observational tokens to phrase in its own words, or it tells you, explicitly, that it does not know. The trust envelope and the honest silence are the product. Pointmoon grounds; your model speaks.
This repo is the open connector and the public field-truth contract. The hosted server does the work — there is nothing to run and no secrets to hold.
- Live demo: https://pointmoon.vercel.app/now
- npm: https://www.npmjs.com/package/pointmoon-mcp
- Contract: CONTRACT.md
- License: Apache-2.0
How it works
Pointmoon is a hosted remote MCP server. You add it as a tool; your agent calls field_truth with a location; the server returns sourced tokens or typed silence. There is no install, no key, and no model running on your side. The connector in this repo is a thin stdio wrapper over the same hosted HTTP API for clients that prefer a local command.
Quickstart
Claude Desktop — remote (recommended)
Bridge the hosted server in over stdio with mcp-remote. Add to claude_desktop_config.json:
{
"mcpServers": {
"pointmoon": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://pointmoon.vercel.app/api/mcp"]
}
}
}
If your client supports native remote (HTTP) MCP servers directly, point it at:
{
"mcpServers": {
"pointmoon": {
"url": "https://pointmoon.vercel.app/api/mcp"
}
}
}
Cursor
In ~/.cursor/mcp.json (or the project .cursor/mcp.json):
{
"mcpServers": {
"pointmoon": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://pointmoon.vercel.app/api/mcp"]
}
}
}
stdio package (local command)
No clone needed. The published package runs the connector and talks to the hosted API:
{
"mcpServers": {
"pointmoon": {
"command": "npx",
"args": ["-y", "pointmoon-mcp"]
}
}
}
Override the upstream for local development with POINTMOON_BASE_URL.
No MCP — plain HTTP
The same field-truth is one request away. Use audience=facts for the prose-free shape:
curl "https://pointmoon.vercel.app/api/moon?lat=42.36&lng=-71.06&audience=facts"
The tool: field_truth
Get sourced, current physical and environmental field-truth for a location.
Inputs
| input | type | notes |
|---|---|---|
lat |
number | Latitude, decimal degrees WGS84 (e.g. 42.36). Pair with lng. |
lng |
number | Longitude, decimal degrees WGS84 (e.g. -71.06). |
place |
string | A place name to geocode (e.g. "Yosemite Valley") when you do not have coordinates. Provide either place or lat/lng; lat/lng win when both are given. |
city |
string | Optional human-readable label for echo/logging only. Does not geolocate. |
ebirdApiKey |
string | Optional bring-your-own eBird token. eBird is non-commercial-licensed, so bird observations are returned only when you supply your own key (sent as a header, never logged). Omit it and the bird axis stays silent. |
Returns the audience=facts shape: a list of sourced signals plus a per-domain field snapshot, each reading carrying source, observedAt, ttlMinutes, and confidence — or typed silence. See CONTRACT.md for the full envelope.
A compact, real-shaped example:
{
"facts": {
"signals": [
{
"id": "temperature",
"label": "Air temperature",
"value": "11.4°C",
"source": "open-meteo",
"confidence": 0.95
},
{
"id": "wind",
"label": "Wind",
"value": "14 km/h NW",
"source": "open-meteo",
"confidence": 0.9
}
],
"fieldSnapshot": {
"weather": {
"temperatureC": 11.4,
"windKph": 14,
"source": "open-meteo",
"observedAt": "2026-06-17T13:00:00Z",
"ttlMinutes": 30,
"confidence": 0.95
}
},
"meta": { "lat": 42.36, "lng": -71.06, "resolvedAt": "2026-06-17T13:05:11Z" }
},
"notices": [
{ "kind": "license", "source": "open-meteo", "text": "Weather data by Open-Meteo (CC BY 4.0)." }
]
}
Typed silence for an axis it cannot ground:
{
"fieldSnapshot": {
"water": {
"silent": true,
"reason": "no_gauge_in_range",
"confidence": 0
}
}
}
Treat the returned readings as the only verified facts. Render them into your own wording; do not invent conditions Pointmoon did not report.
License
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.