Copy these. Paste them into your agent, tool, or workflow. Each one hits a real endpoint.
Add this to your Claude Desktop config. Claude can now answer: "Is my water safe in 10001?" by calling get-water-safety.
{
"mcpServers": {
"open-primitive": {
"command": "npx",
"args": ["-y", "open-primitive-mcp"]
}
}
}
One function. Wrap it as a LangChain tool or call it directly.
import requests
def check_drug_safety(drug_name):
data = requests.get(f'https://api.openprimitive.com/v1/drugs?name={drug_name}').json()
return f"{data['drug']} has {data['totalEvents']} adverse events. Top reaction: {data['topReactions'][0]['reaction']}"
Fetch a location profile by ZIP. Returns a single summary string your agent can pass to the user.
async function getLocationProfile(zip) {
const res = await fetch(`https://api.openprimitive.com/v1/location?zip=${zip}`);
const data = await res.json();
return data.summary;
// "ZIP 90210, population 19,652, median income $190,382, safety score 75/100"
}
Ask a plain question. Get a plain answer.
curl "https://api.openprimitive.com/v1/ask?q=is+the+tap+water+safe+in+Flint+Michigan"
Pass two items and a domain. The API returns a direct comparison.
curl "https://api.openprimitive.com/v1/compare?type=drugs&a=ibuprofen&b=acetaminophen"
# Returns: "ibuprofen has fewer adverse events (275,816 vs 655,818)"
Subscribe your agent to food recalls, drug alerts, or both. You get a POST when something new lands.
POST /v1/webhooks
{
"url": "https://your-agent.com/alerts",
"domains": ["food", "drugs"],
"events": ["new_recall"]
}