Documentation

Agent Integration

Integrate eir-open medication lookups and health data into Claude, LangChain, MCP, or any AI agent framework.

eir-open tools expose a consistent JavaScript API and CLI that any agent framework can use as tools or subprocesses. Whether you’re building with Claude’s tool use, LangChain, OpenClaw, or a custom pipeline — the same underlying packages work everywhere.

Supported Frameworks

Claude Tool Use

Define medication lookups as Claude tools using the Anthropic SDK’s function calling interface.

OpenClaw (Native)

The native integration path — install skills with a single command. See the OpenClaw Integration guide.

LangChain

Wrap the CLI or npm API as a LangChain Tool (Python) or DynamicTool (JS) in your chain or agent.

Generic Subprocess

Any agent framework can shell out to the CLI tools — a simple, universal integration path.

Claude Tool Use

Define eir-open lookups as Claude tools using the Anthropic SDK.

import Anthropic from "@anthropic-ai/sdk";
import { lookupMedication, searchMedications } from "swedish-medications";

const client = new Anthropic();

const tools: Anthropic.Tool[] = [
  {
    name: "lookup_medication",
    description:
      "Look up a Swedish medication by brand name or generic substance name. Returns dosage, side effects, OTC status, and FASS link.",
    input_schema: {
      type: "object",
      properties: {
        query: {
          type: "string",
          description: "Medication brand name or active substance (e.g. 'Alvedon', 'paracetamol')",
        },
      },
      required: ["query"],
    },
  },
  {
    name: "search_medications",
    description:
      "Search across all 9,064 Swedish medications. Returns up to 10 matches with names, substances, and OTC status.",
    input_schema: {
      type: "object",
      properties: {
        query: {
          type: "string",
          description: "Search term for medication name or substance",
        },
      },
      required: ["query"],
    },
  },
];

async function handleToolCall(name: string, input: Record<string, string>) {
  if (name === "lookup_medication") {
    return await lookupMedication(input.query);
  }
  if (name === "search_medications") {
    return await searchMedications(input.query);
  }
}

// Use in a conversation loop
const response = await client.messages.create({
  model: "claude-opus-4-6",
  max_tokens: 1024,
  tools,
  messages: [{ role: "user", content: "What is Alvedon used for?" }],
});

For US medications, swap in us-medications and adjust the tool descriptions accordingly.

LangChain

Python

Wrap the fass-lookup CLI as a LangChain Tool:

import subprocess
from langchain.tools import Tool

def fass_lookup(query: str) -> str:
    result = subprocess.run(
        ["fass-lookup", query],
        capture_output=True, text=True
    )
    return result.stdout or result.stderr

medication_tool = Tool(
    name="swedish_medication_lookup",
    func=fass_lookup,
    description="Look up a Swedish medication by brand or substance name using the FASS database."
)

JavaScript / TypeScript

Use the npm API directly as a LangChain DynamicTool:

import { DynamicTool } from "@langchain/core/tools";
import { lookupMedication } from "swedish-medications";

const medicationTool = new DynamicTool({
  name: "swedish_medication_lookup",
  description:
    "Look up a Swedish medication by brand or substance name. Returns dosage, warnings, and FASS link.",
  func: async (query: string) => {
    const result = await lookupMedication(query);
    return JSON.stringify(result);
  },
});

OpenClaw (Native Skills)

OpenClaw is the native integration path for eir-open medication tools. Skills are installed with a single command and automatically available to any OpenClaw-compatible agent.

See the OpenClaw Integration guide for setup instructions.

MCP (Model Context Protocol)

The EIR MCP Server lets agents load .eir health data files and query them with natural language — no custom tool definitions required. Compatible with Claude Desktop and any MCP client.

See the MCP Server guide for setup and configuration.

Generic Subprocess Pattern

Any agent framework can shell out to the CLI tools. This works in any language with subprocess support.

# Shell
fass-lookup "metformin"
us-medications "lisinopril"
us-medications --interactions "warfarin,aspirin"

# Python
import subprocess, json

result = subprocess.run(
    ["us-medications", "--json", "lisinopril"],
    capture_output=True, text=True
)
data = json.loads(result.stdout)

// Node.js / TypeScript
import { execSync } from "child_process";

const output = execSync(`fass-lookup "paracetamol"`, { encoding: "utf8" });

What Can Agents Do?

Medication Lookup

Brand-to-substance resolution, dosage, warnings, and OTC status

Drug Interactions

Check interactions across the 81,212-medication US FDA database

Multi-Result Search

Search by partial name, substance, or category across both databases

Health Data Parsing

Load and query .eir health records via the MCP Server